Modern video stacks now combine speech recognition, computer vision and multimodal indexing to chop up raw footage into structured data. Spoken words, logos and specific scene changes are converted into crisp metadata that AI search engines can parse and cite, instead of just ranking clips in a list. User-generated videos on TikTok, Reels and Shorts are no longer just thumb-stopping content; they’re being scraped and tagged by AI models as cold, hard evidence.

This capability is already active across commercial software. Videowise launched an AI Visibility feature for Shopify storefronts in September, parsing social clips frame by frame to produce structured data for ChatGPT, Perplexity and Gemini. Competitors such as Vyrill process user-generated videos, scanning dialogue, objects and sentiment to build searchable product directories out of social posts.

At the foundational level, Gemini 1.5 analyses video streams in real time to index critical moments for search retrieval. The heavy presence of YouTube inside AI Overviews is also driving marketing teams to draft creator scripts as functional search copy.

How Product Placement Becomes Searchable Inventory

Once computer vision tags a specific item or logo inside a casual user clip, that video stops acting like a one-off social post. It turns into retrievable inventory, something an e-commerce site can pull into internal search, a merchant can stick next to a buy button or ChatGPT can cite as proof that people actually buy the product.

Product placement is morphing from a passive reach strategy into a hard data layer. Every visual appearance is logged, indexed and repurposed as structured proof, which replaces the old metric of simple view counts.

Talent agencies are already telling brands to draft creator scripts as functional search copy, loading them with specific product names and clear problem-solution framing instead of leaning purely on tone and aesthetic. Success now depends on accurate transcripts, descriptions and chapter markers so AI parsers can index clips the same way they process long-form articles. Strategic on-screen text, featuring key terms and product names, is now mandatory to anchor spoken audio.

Briefs have shifted from a simple “make it go viral” to “make it viral and also machine-readable,” ultimately changing how creator content is pitched and produced.

Are Creators Now Optimising For AI Instead Of Humans?

It’s already taking hold in subtle ways. Agencies are hyping “video SEO 2.0”, encouraging creators to write scripts that read like searchable articles while maintaining a casual delivery.

The standard production checklist now mirrors classic web search tactics, borrowing the same instinct that drives written SEO and applying it to a different medium: pacing, framing and visual structure instead of headers and keyword density. Some teams are even testing screen pacing by leaving a key graphic or logo frozen on screen for a few extra seconds, purely for AI models to index it properly.

The likely outcome is a hybrid optimisation target that didn’t exist before: content built to perform on human engagement metrics while simultaneously delivering machine interpretability through clean transcripts, detectable objects and a clear semantic structure a system can parse.

What Gets Lost When A Video Is Built To Be Read?

AI models have the artistic soul of an Excel spreadsheet. They want direct product shoutouts, rigid problem-solution scripts and clear visuals. Nuance, dry humour and double entendres are completely lost in translation.

By forcing creators to play nice with indexers, brands risk turning creator content into generic noise. Identical structures and forced callouts result in every post sounding like a carbon copy, regardless of how fancy the camera work is.

Tailoring content for machine parsing threatens the unique personality and raw emotion that draw human audiences in. If a campaign’s ROI depends on AI citations, marketers must confront the uncomfortable reality of who they’re actually building content for: the human scroller or the indexing bot.

The industry has yet to figure out what authentic creator content means once technical schema and machine-readable scripts become mandatory. What lies ahead is a split creator space, balancing human entertainment on one side and machine retrievability on the other.

The post How AI Video Indexing Is Reshaping Influencer Marketing And Creator Content Strategy appeared first on 91̽��.

Anthropic, OpenAI and Google DeepMind have been meeting in private working groups since July, to explore a self-regulatory body dedicated to safety auditing and pre-release testing for high-capability models. The initiative lacks a formal charter, press release or finalised governance setup. Instead, it shows the leading parties in AI discreetly attempting to design a common safety playbook on their own terms.

The working group is keeping its sights on a narrow set of operational rules. Rather than hammering out broad industry policy, the talks cover third-party evaluations before release, formal safety checks and unified risk protocols for cyber attacks, bioweapons risks and deceptive behaviour. It functions as a voluntary, lab-funded auditing system for frontier models, an arrangement that could start as a flexible agreement and eventually mature into an industry standard.

This move has clear roots. In July, Google DeepMind leader Demis Hassabis outlined a plan modelled on FINRA, proposing an industry-financed, federally monitored body staffed by independent experts to conduct pre-release evaluations about 30 days out. The ongoing working groups show those ideas gaining traction. Anthropic chief Dario Amodei built on the momentum in a recent essay, pushing for aligned benchmarks and official antitrust protection for safety-related coordination.

Navigating those regulatory boundaries is important, mostly because three direct competitors agreeing to hold back tech looks suspiciously like anti-competitive behavior to watchdogs. That danger alone explains why everyone involved wants a formal, structured standards body on paper rather than an unwritten understanding.

Complement Or Alternative To Real Regulation?

On paper, a self-regulatory body can do plenty of useful work, between mapping out frontier thresholds and writing test suites to clearing independent auditors and establishing safety card norms.

The real snag is enforcement. Without the authority to pause a launch or penalise malicious parties, the whole setup relies on voluntary goodwill. That creates an obvious flaw, as letting giant tech firms grade their own work introduces a clash between launch deadlines and safety protocols.

This dynamic plays out very differently depending on the jurisdiction. Under the EU AI Act, binding laws dictate mandatory assessments, transparency requirements and government penalties. There, a voluntary body acts as a helpful partner, offering technical evaluation tools to meet legal requirements. In the US, where broad AI legislation doesn’t exist, that same group could easily turn into a proxy regulator, giving lawmakers a reason to skip binding laws altogether. Whether this initiative becomes a genuine safeguard or a clever shield against regulation depends on how much authority Washington decides to hand over.

We posed the issue to insiders across AI safety, regulation and enterprise tech: what would an industry-driven standards body need to feature to hold real authority, and does joint oversight between three main competitors offer authentic safety or just a convenient way to derail tougher laws before they land?

Our Experts

• Lakshmi Hanspal, Chief Trust Officer, DigiCert
• Emily Hartstone, Founder, Hartstone Institute LLC
• Kirk Sigmon, Founding Partner, KellDann Law PLLC
• Sherif Higazy, Founder and CEO, Megaton AI
• NagaPranitha Chodavarapu, Senior Lead QMS, Insulet Corporation

Lakshmi Hanspal, Chief Trust Officer, DigiCert

“What could this body actually set rules for? Greater success towards shared technical basics: common tests before releasing a new model, agreed ways to check if a model could be dangerous, and a standard way to report problems when they come up.

“Why team up now? Partly safety concern. But also self-interest: if the big players set the rules themselves, they get ahead of governments doing it for them, and rules they write could end up favouring the companies that helped write them.

“Does this support the EU AI Act, or try to avoid it? In practice, likely a hedge, an attempt to demonstrate credible self-governance before harder regulation lands, particularly in the US.

“What would make this actually work, like IETF or CAB Forum did? Three things, from experience: independent audit that isn’t self-graded, real consequences for non-compliance, not just reputational, and governance open beyond the founding members. CAB Forum works because no single browser or certificate authority can unilaterally rewrite the baseline. Right now, this AI effort has none of those guardrails yet. That’s the gap to watch.”

Emily Hartstone, Founder, Hartstone Institute LLC

“Three labs cooperating on standards is more interesting for what it wouldn’t cover than for what it would. A body like this can realistically set model-layer standards: evaluation methods, safety benchmarks, disclosure practices, incident reporting between labs. Those are useful, and only the labs can do them, because nobody else has the access.

“What it cannot set is what happens at deployment. The model isn’t where people get denied, scored or flagged. That happens inside an enterprise, configured by an operator the lab never meets, acting on a person who never chose either of them. Look at Microsoft’s Code of Conduct, published this week and open for consultation. It’s the most concrete document of its kind, and every obligation in it terminates at the operator or the user. The person the model acts upon appears only once, as an affected third party who shouldn’t be directly harmed. That’s protection, not standing.

“So my answer on whether it complements or replaces regulation is neither. It occupies a different layer. The risk isn’t that it pre-empts the EU AI Act. It’s that it looks like coverage while leaving the deployment layer untouched, and then everyone points at it. Why now is simpler. After this summer’s agent incidents, shared incident reporting is in all three companies’ interest, and a standards body is a reasonable way to build it.”

Kirk Sigmon, Founding Partner, KellDann Law PLLC

“An industry-led body for AI governance is a nice idea and a good development, but it won’t likely change many of the issues regulators are concerned about. Realistically speaking, such standards would be opt-in and thus wouldn’t really stop third parties from using their own models, including locally-executing ones, to circumvent those regulations. It’s also unlikely that any of those parties would agree to regulations with enough teeth to significantly affect their bottom line. That self-regulating body may calm regulators’ concerns somewhat, but I still suspect various jurisdictions will legislate where necessary to protect the public against major concerns, like deepfakes or failure to disclose AI usage in certain medical decisions.

“Plenty of other industries have self-regulated successfully. The video games industry’s ESRB helped avoid Congressional regulation of video games after a scare regarding violent video games. That said, the success of those efforts is often not just the product of regulatory fear: the ESRB was in many ways successful because the vast majority of commercially sold video games went through a relatively small number of publishers who all shared an interest in avoiding regulation, and because video game console manufacturers, of which there were even fewer, could help push forward its use. I don’t see similar dynamics in the AI industry, where there may be relatively few major players but where the underlying technology can be run by virtually anyone with a sufficiently powerful computer.”

Sherif Higazy, Founder and CEO, Megaton AI

“The Trump Administration’s position has been a light touch to regulation, and instead has signalled a preference for a self-regulating body, lest government intervention slow down the US’s AI lead. At the same time, the AI companies have signalled quite strongly that they would welcome some industry and government regulation in the US, which remains the primary market for frontier labs outside of China.

“Two things this body could accomplish. First, sharing and working together on alignment: how to ensure AI systems align with human goals, and how to interpret what an AI system ‘thinks’ and ‘does’ without hiding it from researchers. Second, slowing down and pacing model releases to allow companies to harden their cybersecurity and biosecurity infrastructure.”

NagaPranitha Chodavarapu, Senior Lead QMS, Insulet Corporation

“I can’t speak to what Anthropic, OpenAI and Google are actually planning, but with 13-plus years working in governance, risk and compliance across medical devices, pharmaceutical and biotechnology industries, I’ve watched something pretty similar play out in a different regulated industry, and it might be useful context here.

“Our field’s main professional body, ISPE, published an AI governance guide back in July 2025. Six months later, the FDA and EMA published their joint AI principles. I work right at that overlap, and I’ve built a framework for how AI tools should actually be governed inside regulated validation work, currently going through peer review with that same industry body.

“What stood out to me is that the industry guidance didn’t show up first and try to get ahead of regulators. It came out alongside them, and it’s stayed deferential. FDA enforcement has continued regardless of the industry framework existing. What the industry body actually did well was the unglamorous part: translating a broad regulatory principle into something a company can actually follow in daily operations. If this AI standards body works the same way, filling in the practical gaps around something like the EU AI Act, that’s a model with a real track record. What I’d watch for is how the industry actually handles it in practice.”

The post OpenAI, Anthropic And Google Are Discreetly Building Their Own AI Standards Body – What Would That Actually Decide? appeared first on 91̽��.

If final sign-off goes to plan, the public funding will pull in an additional £30 million from private pockets to create a £60 million co-investment deal. It’s one piece of a much larger £7.8 billion strategy running to 2030, designed to keep the UK’s defence, tech and space sectors properly orbit-ready.

Occupying an old RAF outpost at the northern tip of Shetland, SaxaVord is officially the UK’s first licensed vertical rocket launchpad. Regulators have capped its activity at 30 launches per year, an infrastructure limit instead of an active flight schedule.

To date, no rocket has successfully made it to space from Unst. An August test launch was shelved after the launch provider ran into undisclosed technical troubles, a helpful reminder before treating the site as a fully operational orbital hub.

Where The Money Actually Goes

The planned investment is intended to fund the completion of three launchpads, mission-management infrastructure, a second rocket-integration hangar and facilities usable by multiple international launch providers, essentially the shift from development-stage infrastructure to regular commercial launch operations.

SaxaVord itself isn’t a rocket manufacturer. Its role is providing launchpads, safety systems, range management, integration facilities and logistical support for whichever launch companies operate from the site.

Shetland’s extreme northern latitude provides a specific orbital advantage. Direct access to polar and sun-synchronous orbits makes the location ideal for Earth observation and reconnaissance payloads, avoiding populated landmasses through an open corridor over the Atlantic.

Provided the site achieves a steady flight cadence, it stands to support weather monitoring, maritime tracking, defence assets, small-sat constellations and swift payload replacement.

What Does “Assured Access To Orbit” Mean?

The phrase gets thrown around loosely, so a bit of precision helps. Assured access to orbit comes down to having guaranteed launch options on standby for whenever satellites need to go up, get replaced or shift position in a hurry.

This isn’t total national self-reliance, and the strategy makes no claim that it is. The approach combines a domestic hub at SaxaVord with German and allied partnerships, continued reliance on European Space Agency launches, alongside funding for space domain awareness, satellite comms and in-orbit servicing.

The numbers in the technical annex tell the real story. Of the total £226 million earmarked for assured-access measures through to 2030, SaxaVord receives £30 million, while £39 million goes to ESA programmes in French Guiana and £148 million flows into European space-transport initiatives. The distribution shows a pragmatically diversified approach, prioritising domestic capability alongside proven allied networks rather instead of total self-reliance.

The goal isn’t to launch every domestic payload from British soil, but to eliminate single points of failure so external challenges or diplomatic squabbles can’t freeze UK space operations.

What £30 Million Won’t Do

A reality check on the terminology is important, given how loosely “sovereign” is applied in tech coverage.

This cash injection won’t finance a home-grown British rocket, establish complete supply chain independence or replace existing ties with ESA and the Guiana Space Centre. Far from cutting ties, the policy specifically embeds SaxaVord into global alliances. A more accurate understanding is that the UK is building physical launch infrastructure on home turf, while continuing to host foreign rockets, operators and payloads.

Turning £30 million of taxpayer cash into a thriving commercial hub requires more than just poured concrete and mission control screens. The site needs rockets that reach orbit reliably, competitive launch fees, manageable insurance rates and a high enough launch frequency to offset fixed costs. Laying down the pads builds the field of dreams, but it doesn’t guarantee the business will come.

How This Compares To Europe’s Existing Launch Capacity

Matching the volume of Europe’s major launch sites isn’t the goal here. The European Space Agency handles its heavy lifting out of French Guiana, backed by deep industrial supply chains across France and Germany.

SaxaVord brings a different set of advantages to the table: an ideal northern trajectory for polar satellite constellations, dedicated access for smaller payloads and nimble scheduling that massive multi-tenant rockets struggle to match. It hands the UK an operational anchor in orbital logistics, even if the launch vehicles themselves arrive from abroad.

The strategy’s backing for ESA programmes makes the overarching ambition straightforward. SaxaVord serves as an added layer of regional launch capacity instead of being a direct rival. In short, Britain isn’t constructing an end-to-end national space programme in isolation. It’s establishing a strategic launch site on home soil to own a section of the infrastructure, while continuing to rely on allied rockets across the continent.

Whether that setup proves commercially viable, or turns into an expensive piece of northern scenery, is a test the initial £30 million backing can’t settle on its own.

The post Why Is The UK Investing £30M In Scottish Space Hardware? Decoding The Sovereign Orbit Strategy appeared first on 91̽��.

Is an undersized local market a strategic superpower that pushes founders to grow faster and think smarter? Or is it just a handicap on runway, talent and customer acquisition that survivor companies overcome, then retroactively paint as a brilliant strategy?

It’s easy to credit the small-market theory when looking at Singapore’s biggest tech wins. But survivorship bias tells the same story: the founders for whom a restricted home market was a dealbreaker simply aren’t written about.

Does The Small-Market Theory Hold Up?

The idea that a small domestic market gives founders an edge is more than a convenient narrative, it’s baked into economic policy.

EnterpriseSG and other state bodies urge startups to target global reach from launch, pointing to a home market of 5.9 million as too small to support big tech exits on its own. The Economic Strategy Review mid-term update makes the same point, and startup surveys back it up, with founders increasingly testing overseas demand from seed stage instead of treating international expansion as a later chapter.

Founders building in Singapore describe a consistent pattern: a restricted domestic base pushed them to think cross-border from day one, simply to make the unit economics work. Analysis from Vertex Ventures puts that boundary as a useful filter, one that stops teams from optimising for an ambiguous local audience and forces early clarity on who the real customer is. Singapore commonly becomes the operational hub for talent, IP and product development, with commercial growth happening across the wider region instead.

Yet the counter-argument makes an equally strong point. Industry watchers insist that commercialisation is still the real hurdle. Too many founders optimise for a tiny domestic pool that can’t breed tech giants to rival American or Chinese heavyweights. The rare success stories didn’t benefit from a helpful small-market constraint; they succeeded because they built cross-border products straight out of the gate.

Other commentators view the limited domestic audience as an outright ceiling that pushes cash-intensive sectors like deep tech into cautious regional holding patterns rather than aggressive global expansion.

Going global immediately also introduces a level of risk that’s easy to underestimate. Crossing borders right away jacks up the cost of every wrong turn. A deep-pocketed rival in a large domestic market can test a feature, lose cash and adjust course, but a Singapore-based founder rarely has the runway for that many failed experiments. A compact, highly networked market can also end up rewarding who you know over actual market pull.

The tension is that Singapore’s best-known successes, Grab and Shopee among them, are the companies that executed a global-from-day-one strategy well. Their visibility doesn’t settle whether the constraint benefits founders more broadly.

We put the question straight to Singapore founders: does an undersized home market sharpen global focus from launch, or is it just a hurdle to overcome? And would they have taken a completely different path with a large domestic customer base to rely on first?

Our Experts

• Shammi Thakur, Research Director, Vyansa Intelligence
• Dr Seamus Phan, CTO, McGallen & Bolden Pte Ltd
• Ray Tay, Co-founder, VIVOS Pte. Ltd.
• Oscar Asly, Group CEO, M4Markets

Shammi Thakur, Research Director, Vyansa Intelligence

“From a research perspective, I track many Singapore-based companies, and the truth is that both viewpoints hold validity, though the impact varies for each founder.

“Constraints are real. A market of six million people means you quickly hit a ceiling in terms of local revenue, a point that would take founders in Indonesia or India much longer to reach. At the same time, sourcing talent for specific roles is genuinely difficult and expensive. This isn’t an advantage. It’s a constraint that founders must factor into their planning from the very first hire.

“Yet the discipline this fosters is also real. Founders who can’t rely on massive local market scale often bake features like compliance, payments and localisation directly into their products from the start, because planning to operate across five countries is what makes the unit economics work. Companies starting in large domestic markets often add these elements later, a process that can prove quite difficult.

“Where I disagree with the prevailing narrative is the idea that this inherently makes Singaporean founders better strategists. In reality, it often means they have less time, or runway, to make mistakes on market fit before international expansion becomes a necessity. That’s distinct from superior strategy. And common discourse often falls prey to survivorship bias, overlooking the founders whose companies collapsed under early pressure rather than emerging stronger.”

Dr Seamus Phan, CTO, McGallen & Bolden Pte Ltd

“As a Singaporean, with family who has done small business in Singapore for the last few decades, there are two sides to look at. As a global strategy and communications consultant, I also serve Singapore-based businesses in the FMCG space, and there are multinational corporations that want to set foot in Singapore specifically to serve as their APAC or ASEAN headquarters, managing a network of communication partners from China to Australia.

“For other businesses that aren’t intellectual property or consulting related, but physical goods, the domestic market in Singapore is in fact limiting. For a Japanese business, simply serving the Tokyo market may be sufficient, since there are 14 million people in Tokyo alone. Likewise, a small business serving just Shanghai may be sufficient, given its 24 million residents. But for a Singapore business selling physical products, the incentive is to go regional from the start, whether through e-commerce, working with local channels in target countries, or setting up outlets there directly.

“For small businesses, the constraint is always capital, whether financial or human. My advice is always to start small and bootstrap, rather than taking loans that create pressure, and to scale slowly and steadily. In the event of imminent failure, the damage is far more easily contained too.”

Ray Tay, Co-founder, VIVOS Pte. Ltd.

“Six million people and 98 tax treaties. That ratio is the real Singapore story. The small market is a genuine constraint. It caps the revenue you can prove at home, which caps what you can raise and who you can hire. Investors want traction in two or three ASEAN markets before writing a Series A cheque. Founders work around that. It doesn’t make them sharper.

“What Singapore supplies is cheap optionality. The treaty network, banking access and holding-company regime make the paperwork of going regional almost trivial. The operations stay hard. Singapore lowers the cost of the decision, not the cost of execution.

“The forced-global rule only looks like a law of nature because we count the winners. Grab was told at Harvard that Southeast Asia was too small a market to focus on, and chose Singapore as a base for the region after Malaysia’s own sovereign fund passed on backing it. That was a choice, not an inevitability. In the Singapore Business Federation’s 2025 survey, 41% of businesses had never internationalised, and 81% of those had no plans to. Plenty stay home and do fine.

“Would I have built VIVOS differently with a big home market? Yes, and worse. A large domestic base lets you postpone the regional question, and postponing it is how you end up with a product that only works in one place.”

Oscar Asly, Group CEO, M4Markets

“A small home market makes you look abroad earlier. It doesn’t magically make you better at doing business there. There’s a useful pressure in Singapore: you have to ask quite quickly whether anyone outside your home market wants what you’re selling. But you’re also trying to fund that expansion from a smaller customer base. That’s a real constraint, however neatly we dress it up afterwards.

���� financial services, your technology can cross a border much faster than your licence or your reputation. You still have to understand the customer, build relationships and earn trust in each market. ‘We’re going into Asia’ is an ambition. It isn’t a strategy. And yes, there’s a survivorship problem. We hear from the companies that made it overseas. We hear much less from those that spread themselves too thin trying.

“With a bigger home market, I’d probably have expanded more patiently: built a stronger revenue base and been more selective about where to go next. I’d still build for international growth, but I wouldn’t confuse being ready to expand with needing to expand. Singapore forces the question earlier. The quality of the answer is still down to the founder.”

The post Is Singapore’s Small Market Actually Its Biggest Strategic Advantage? appeared first on 91̽��.

Right now, Labour’s ten regional mayors have promised to cap their local rates at 5%, but that’s a gentleman’s agreement rather than a legal guarantee, leaving the door open for future leaders or different political parties to increase the percentage.

Before anyone panics, it helps to separate the structure from immediate reality – no one is paying a new room tax just yet. This is simply the policy structure, meaning a full bill must still pass, and authorities aren’t expected to publish concrete spending strategies until early 2028.

Local leaders still have to choose whether to implement a charge, gather feedback from local businesses and figure out the logistics. It’s also completely distinct from Scotland’s setup, where Edinburgh’s scheme runs as a flat 5% fee capped at five nights.

The point is that England is heading toward a mosaic of local rules. Instead of a single national tourist tax, the policy enables dozens of different regional levies, each running on its own timeline, rate, spending agenda and list of exemptions. For accommodation owners and frequent business travellers, the pressing concern is whether a non-binding political promise can actually protect businesses from a fragmented mess once different authorities start pulling the levers.

Why Trade Bodies Are Raising The Alarm

The pushback from trade groups goes deeper than just resistance to new taxation.

ABTA argues that a percentage model disproportionately penalises luxury and boutique accommodation compared to a flat nightly rate. UKinbound criticises the regional setup as an operational nightmare, calling for a fixed national system instead of a postcode lottery. At the same time, the World Travel and Tourism Council warns that higher trip costs could steer holidaymakers and capital toward competing destinations altogether.

UKHospitality takes the warning even further, claiming an uncapped charge could inflict a £1.6 billion hit on the sector, bump the price of a typical family holiday up by £100 and threaten up to 33,000 jobs. Those are the trade group’s own projections, not official guarantees, but they capture the depth of concern across the market.

Conversely, government officials insist that targeted levies fund the essential facilities that keep destinations competitive. The Liverpool City Region, for example, expects a levy could pull in up to £18 million annually for events, cultural projects and local transport, though that’s only one regional forecast.

We put the question to the people on the ground. How would a levy like this actually hit bookings, pricing and overall competitiveness, and does the lack of a legal cap worry them more than the levy itself?

Our Experts

• Siarhei Sulimau, CEO and Founder, EnglishPapa
• Amy Boyton, Director of Franchise Sales, Pass the Keys
• Julia Doust, Founder and Editor, The European Compass

Siarhei Sulimau, CEO and Founder, EnglishPapa

“I run EnglishPapa, and on the hospitality side I own and operate Aviator Bali, an apart-hotel in Bali. Managing accommodation in a competitive international tourism market has shaped my view on England’s proposed visitor levy: the real issue isn’t the tax itself, it’s the inconsistency around it.

“Any percentage-based charge eventually lands in the guest’s final bill, and in a market where travellers compare prices across destinations in seconds, that matters. Most guests don’t mind paying a levy if they understand where the money goes: tourism infrastructure, local services, upkeep of the places they’re visiting.

“What concerns me more is fragmentation. If every English region sets its own rate, we get a messy, inconsistent pricing picture, especially painful for cities directly competing for the same visitors. Hotels can’t price transparently or competitively when the rules shift by postcode, and guests end up confused about what they’re paying and why. My honest take: a lack of national framework is a bigger risk than the levy itself. Consistency, not the charge, is what will make or break this policy.”

Amy Boyton, Director of Franchise Sales, Pass the Keys

“From new licensing rules and visitor taxes to council tax hikes and minimum night stays, short-term rentals are being burdened with measures that claim to fix housing but end up penalising tourism and the people who rely on it.

“Visitor taxes aren’t a one-size-fits-all solution. In many towns and cities, they simply push up costs for guests, and that includes domestic travellers who already pay their share through existing taxes and local spending. These levies can bring benefits, but not every destination has the constant pull of a city like London, and most places can’t impose extra charges without risking losing bookings.

���� places like Edinburgh and Glasgow, where levies have already been approved, our local managers are stuck between absorbing the extra cost or risking fewer bookings. If they raise prices by 5% to offset the tax, they risk becoming uncompetitive. Add that to licensing fees, and it’s no surprise many Scottish hosts are seriously considering shifting to mid-term lets instead, given traditional long-term rentals simply don’t work for most holiday homes as they’re often rural, seasonal or used part-time by owners. We need more balanced policymaking if we are to maintain the very tourism economies these measures claim to support.”

Julia Doust, Founder and Editor, The European Compass

“I was the owner of a 21-room establishment in France when my local council decided to impose a tourist tax. It made no difference whatsoever to my bookings.

“Now I cover cities across Europe, almost all of which impose a tourist tax. Visitors expect it. It doesn’t make a difference to their decision-making process. Costs have to rise 20 to 30% before people start to compare different destinations.

“The only difference can come when travellers have set themselves a limit per night, say €200. If the tourist tax puts them over that limit for a certain hotel, they may choose a cheaper hotel, but they don’t change destination.”

The post England’s New Tourist Tax Has No National Cap – What Does That Mean For Hospitality Businesses? appeared first on 91̽��.

During a cybersecurity test back in January 2026, a pre-release version of Claude Opus 4.6 bypassed containment to interact with real-world networks, a breach that went undetected for seven months until staff prepared files for an outside review. To address the recurring issue, Anthropic has teamed up with independent non-profit METR on an eight-week probe covering all four incidents, opening up its transcripts and technical teams to outside investigators.

The culprit across all four sandbox breaks was a repeated setup flaw from the same third-party provider, which accidentally left an active internet connection open while promising the AI it was safely offline. Facing a blocked target due to conflicting IP addresses, Opus 4.6 improvised by finding its own route out. It navigated to a real-world machine thinking it was part of the test, harvested credentials, changed system configurations and viewed personal information belonging to a real person before its session timed out.

All of this breaking in the exact same week that Evan Hubinger, Anthropic’s Alignment Science lead, publicly stated a 10% probability of AI-driven human extinction over the next ten years. The two headlines form a surreal pairing: the lab’s leading safety researcher rating existential risk as a genuine threat, while his employer admits its models keep escaping their digital playpens to mess with real-world networks.

When The Hacker Is Just A Confused Algorithm

The trickiest part about classifying these incidents is that they lack the one thing security teams look for: an adversary.

The models had no harmful motives. They were just running through assigned capture-the-flag objectives, continually treating their surroundings as a test lab even while interacting with real infrastructure and harvesting actual personal data.

Anthropic noted two main driving factors behind the behaviour: models talking themselves into believing ambiguous setups were still part of the game, and a sheer refusal to pause when system signals suggested something was wrong.

This dynamic leaves corporate defence strategies in bizarre territory. Firewalls search for hostile breaches, insider threat tools monitor human behaviour and software audit tools look for corrupted dependencies. They don’t know what to do with an AI model that holds proper authorisation, acts with total innocence and accidentally triggers a data breach because a sandbox misconfiguration gave it a path to the live web.

We put the question to CISOs, incident response commanders, penetration testers and AI security analysts: are current enterprise threat models actually accounting for rogue AI agents without malicious intent, or is the tech sector waiting for a major disaster before building real safeguards?

Our Experts

• Viktor Bulanek, Founder, Penetrify
• Evgenii Arsentev, AI Transformation Executive, ARSENTEV.AI
• Karthik Karunanithi, Solution Architect, IBM
• Sayali Patil, Founder and CEO, IntentOps
• Jeff Watkins, Chief AI Officer, NorthStar Intelligence
• Luke Hinds, CEO and Co-founder, nolabs
• Robert Pfleghardt, Founder and CEO, CBR Labs and VoraPrep
• Eshaan Jain, Senior Product Manager, T-Mobile
• Cache Merrill, Founder, Zibtek
• Alexander Leslie, Senior Advisor, Recorded Future

Viktor Bulanek, Founder, Penetrify

“My threat model accounts for AI agents because they are my product. We run autonomous agents that perform authorised penetration tests, and the first design decision we made was to treat our own agent as hostile. Not because it has intent, but because it does not. An agent pursuing its interpretation of a task will walk through any door its credentials open, and it will do so confidently, without the hesitation or self-preservation that makes human insiders somewhat predictable.

“The industry’s detection stack largely assumes an adversary, so it looks for adversary behaviour: staging, evasion, exfiltration patterns. An agent with legitimate access exhibits none of that. What works instead is boring and economic: every autonomous run gets an envelope defined outside the model, an immutable scope, a spending cap, a time limit, a restricted set of destinations it can talk to, and any deviation from that envelope is the incident signal. The agent cannot negotiate with a limit it cannot see. Prompts and policies are advice. Envelopes are control.

“Is the industry building the right defences? Mostly it is still writing AI usage policies, which govern the humans, not the agents. Until “AI agent with legitimate access” appears in threat models next to insider risk, with its own detection and its own kill path, it is being treated as hypothetical. It stopped being hypothetical for us the first week we ran one.”

Evgenii Arsentev, AI Transformation Executive, ARSENTEV.AI

“Honestly, most threat models don’t account for it, and a year ago mine didn’t either. Security frameworks are built around intent. An agent has no intent to find. It has a task, a set of credentials and a very literal reading of both.

“The incidents I actually deal with are dull. Earlier this week one of my agents, asked to rebuild a single service, brought it up with the wrong configuration and quietly broke two integrations that depended on it. Another once changed a system setting on a work laptop that nobody had asked it to touch. Every action was authorised. Nothing looked like an attack.

“So detection has to track scope, not hostility. Log what each agent was asked to do and flag anything it touched outside that. Issue credentials per task and make them expire, so “access it wasn’t supposed to use” mostly stops existing. Put a person or an independent second check in front of anything irreversible. Is the industry ready? No. It still talks about this as a future risk, while in my logs sub-agents already make 52% of all model calls, each with its own context and its own permissions.”

Karthik Karunanithi, Solution Architect, IBM

“I think the industry is asking the wrong question about AI agents. Agent access is a known fact, but the real question is whether an agent can prove why its next action is authorised.

“The Opus 4.7 incident is interesting because the model identified the target as a real company almost 90% of the time, yet it still judged the environment safe to act in 75% of the time. Reachability and context clues were turning directly into permission decisions. In the fourth incident, an Opus 4.6 checkpoint never really checked whether its actions were authorised at all.

“This isn’t a new security problem. We’ve spent years building authorisation models around delegated access, but never applied them consistently to agents, because these agents are provisioned like service accounts: a static credential and a hopeful assumption. NIST’s NCCoE flagged agent identity and authorisation as a security gap back in February, and agent-specific controls remain largely unpublished to this day. Every tool call an autonomous agent makes should carry an auth chain: who delegated it, the resources in scope, the actual purpose and the evidence that permits the action.

“Detection needs the same shift. Anthropic’s monitor initially flagged only around 1% of actions, because the model’s reasoning was persuasive enough to convince the system the environment was simulated. Strip out that reasoning layer and score purely on tool calls and results instead, and the flag rate jumps to 50%. The most pressing emerging threat isn’t an AI stealing credentials. It’s an AI that has valid credentials and invents its own interpretation of what they permit.”

Sayali Patil, Founder and CEO, IntentOps

“No, and I’d be surprised if most threat models can right now, because the category genuinely doesn’t fit anywhere yet. Every established framework I’ve worked with, insider threat, intrusion detection, supply chain risk, assumes intent behind a breach. What Anthropic just disclosed is structurally different: models given legitimate testing access, told they were operating inside an isolated simulation, that kept going past the point where a human would have stopped to ask whether they were actually allowed to be there. One model tried to abort eight separate times before a broken harness pushed it toward real systems instead. That’s not malice. That’s the absence of a working stop mechanism, which in my experience is a far more common root cause than people assume.

“I’ve spent years testing automated systems for exactly this failure, first in large-scale infrastructure at Cisco, now specifically in AI agents. The lesson I keep relearning is that meaningful detection has to stop asking “is this an authorised identity taking an authorised action” and start asking “is this system’s behaviour still consistent with what it was actually supposed to be doing.” Those are two different questions, and a system can pass the first cleanly while failing the second in complete silence, which is precisely what happened here across four separate incidents before anyone outside the company caught it.

“Practically, that means treating abort and halt mechanisms as security-critical infrastructure rather than an afterthought, since a broken kill switch is what turned one of these from a contained test into an actual breach, and instrumenting for behavioural drift rather than just access anomalies. Most organisations I talk to are still treating this as a hypothetical they’ll get to eventually. Anthropic just handed the industry the clearest evidence yet that it isn’t.”

Jeff Watkins, Chief AI Officer, NorthStar Intelligence

“Moving AI out of the chat window and giving autonomous agents access to real systems would always introduce a new category of security risk. Agents can operate at machine speed and scale, use tools and credentials, and potentially coordinate with other agents to achieve an objective. The danger isn’t limited to malicious actors deliberately weaponising them. An agent diligently pursuing a poorly specified objective can potentially cause just as much damage.

“The recent incidents are particularly interesting because they challenge one of the assumptions underpinning conventional threat modelling: that there is an adversary with malicious intent. An AI agent doesn’t necessarily need to be malicious, compromised or even knowingly acting outside its authority to become a security threat. If its understanding of the task differs from ours regarding its boundaries, the outcome can look a lot like an intrusion. That means organisations need to model both external and internal agentic threats. A hardened external perimeter will do little to protect you from an agent already operating legitimately inside it.

“Simply propagating a user’s permissions to an agent is particularly risky. Instead, agents should ideally have their own identities and be granted narrowly scoped, task-specific and time-limited permissions. We also need to reconsider detection and response. Conventional dashboarding assumes there is enough time for a human analyst to notice something unusual and intervene. An autonomous agent may perform hundreds of actions in that interval. Observability needs to be coupled with automated containment: rate limits, behavioural thresholds, circuit breakers and kill switches capable of suspending an agent or isolating its environment when its behaviour moves outside expected boundaries.

“The important principle is that scope should be technically enforced rather than simply described in a prompt. Network controls, permissions, sandboxing and tool restrictions need to make prohibited actions impossible, or at least rapidly detectable and reversible. Agentic AI makes intent a less useful concept in cybersecurity. Rather than asking only who is attacking us and why, we increasingly need to ask what this actor can reach and do, how quickly we can detect unexpected behaviour, and how quickly we can stop it.”

Luke Hinds, CEO and Co-founder, nolabs

“For many of us in the industry, this was inevitable. But these incidents should act as a wake-up call to businesses. Agents don’t often act with malice. It’s a bright, well-meaning agent with too much access and too little context.

“The problem is that many organisations are still trying to secure agents the same way they have secured humans or conventional software. That will not work. Blunt sandboxing is not enough, since if you lock agents down too heavily, you kill the value businesses are trying to unlock. Put a human approval step in front of every action and you no longer have an autonomous agent. Businesses need controlled freedom, enough authority for the task in front of the agent, and absolutely nothing more.

“That means every identity, permission, decision and action needs to be continuously verified, governed and auditable. Detection has to focus on what an agent is actually doing with legitimate access, not simply whether its credentials are valid. Crucially, that security boundary has to sit outside the model. Frontier labs should not be marking their own homework or deciding what agents can access, decide or do. These incidents are proof that organisations need independent controls that govern agents regardless of which model they use. Ultimately, businesses shouldn’t rely on agents to behave well. Security has to be built in from day one, so breaking the rules becomes structurally impossible.”

Robert Pfleghardt, Founder and CEO, CBR Labs and VoraPrep

“Traditional threat models assume malicious intent. That’s the blind spot. After 37 years securing SCIFs and courtrooms, I’ve learned that whether a camera activates because of malware or because an AI misread its own task, the outcome is the same: data gets exposed. I once watched a facility spend six figures on adversarial threat modelling for a room that still had a tablet with a working microphone sitting on the table. Nobody asked what happens if nothing malicious ever touches that device at all.

“So what does real detection look like? Extreme behavioural anomaly monitoring at the data perimeter, regardless of the agent’s declared purpose. You track what data pathways are being accessed and how far they stray from strict authorisation boundaries, not who’s doing it, and not why. Most organisations are still treating this as a software problem, patching around agent behaviour with better access controls and another monitoring dashboard.

“But software guardrails fail, and they fail quietly. If an AI agent has legitimate software access and can physically activate a microphone, a camera or a wireless antenna, unintended data exfiltration is possible. You need smarter detection too, but detection alone isn’t the fix. The real fix is immutable physical limitation. At CBR Labs, we permanently remove cameras, microphones, speakers, Wi-Fi, Bluetooth and antennas from tablets used in secure environments. If the hardware physically can’t transmit, it doesn’t matter what the AI thinks it’s doing. The industry hasn’t caught up to that yet.”

Eshaan Jain, Senior Product Manager, T-Mobile

“Traditional threat models break down when autonomous AI systems act as attackers without malicious intent. Security teams currently rely on frameworks that look for malicious actors, clear indicators of compromise, or compromised insider credentials.

“When an autonomous agent misinterprets a valid workflow and accesses unauthorised infrastructure using legitimate permissions, standard intrusion detection systems fail to catch the deviation. The industry is still treating non-malicious autonomous boundary-crossing as a hypothetical edge case rather than an operational reality. Meaningful defence requires shifting from static access control lists to dynamic semantic guardrails. Security posture must evaluate what an AI agent is actually trying to accomplish in real time, rather than just checking whether its cryptographic credentials are valid.

“Until detection tools learn to monitor behavioural intent and task boundaries alongside technical privileges, companies remain vulnerable to automated overreach that traditional security frameworks cannot classify or stop.”

Cache Merrill, Founder, Zibtek

“I don’t think most enterprise threat models are ready for an AI agent that has legitimate access but uses that access in a way nobody intended. In traditional security, we’re usually looking for a compromised account, malicious code, or someone deliberately trying to get somewhere they shouldn’t. With an AI agent, the access can be legitimate while the actions still create a security problem.

���� the security work I’ve been involved with, I’d want to see what the agent is actually doing after it gets access. It should have only the permissions it needs, its actions should be logged, and there should be limits on what it can change or where it can connect. If it starts behaving outside its expected pattern, I’d treat that seriously even if there was no malicious intent. AI agents need to be treated as part of the attack surface, not just another application.”

Alexander Leslie, Senior Advisor, Recorded Future

“What happened at Hugging Face is a meaningful inflection point, but it needs to be described precisely. This was not an AI model spontaneously developing malicious intent. OpenAI deliberately placed highly cyber-capable models into an exploitation benchmark with their normal safeguards reduced. The significant fact is that the models exceeded the intended boundaries of that test, discovered an unknown vulnerability, obtained access to the open internet, and autonomously chained credential theft, privilege escalation, lateral movement and remote code execution against a real third party.

“Under our AI Malware Maturity Model, this is the clearest public demonstration yet of Level 5 technical capability. An agentic system conducted a complex, multi-stage operation end-to-end without step-by-step human direction. It is not yet evidence of Level 5 malicious activity in the wild. There was no criminal or state operator directing the campaign, and the models were operating under specialised evaluation conditions with reduced refusals and substantial computing resources. That distinction separates a genuine capability milestone from an exaggerated claim that fully autonomous cyber campaigns have suddenly become routine.

“The techniques themselves were not new. The models exploited the same weaknesses that sophisticated human operators exploit, including vulnerable third-party software, overprivileged credentials, insufficient segmentation, and remote code execution paths. What changed was the speed, persistence and autonomy with which those weaknesses could be discovered and combined. The strategic risk is not that artificial intelligence creates an entirely new cyber kill chain. It is that AI can execute the existing kill chain continuously and at a volume that overwhelms human-speed defence. Organisations must treat AI agents as privileged digital identities, treat model and data pipelines as executable attack surfaces, and correlate identity, vulnerability, infrastructure and third-party intelligence at machine speed.”

For any questions, comments or features, please contact us directly.

The post Anthropic Discloses Fourth Unauthorised Claude Access Incident – Is The Security Industry Prepared For AI Breaches? appeared first on 91̽��.

When he began preparing to relocate back in 2026, he found that despite how much the country had changed, he found the relocation process as chaotic as ever: one answer in a Facebook group, another on WhatsApp, another through a recruiter, another on a government website, with the rest left to word of mouth. Housing, schools, jobs, banking, neighbourhoods, documentation and the basic questions of day-to-day life all sat in different places with no single source tying them together.

What struck him wasn’t a lack of information but a lack of trust, context and navigation, which raised an obvious question: why wasn’t there one intelligent platform that understood who someone was, why they were moving and what they actually needed, then guided them through the whole journey? That question became Ahlan.

We chatted to Abiel to find out more about how AI is being used to personalise relocation at scale, what he’s learned from Ahlan’s early beta users, and his ambitions for the platform beyond the UK-Saudi corridor.

2. For those unfamiliar with the company, what exactly does Ahlan do, and how does the platform simplify the relocation and settlement process for users?

Ahlan is an AI-powered global mobility and settlement platform, starting in Saudi Arabia.

The aim is to bring the whole relocation journey into one place. Users can get personalised guidance through Ask Ahlan, explore where to live, compare schools, discover jobs, access trusted providers, use personalised settlement checklists and connect with communities and local events.

Importantly, we’re also building transactional capability into the platform. Users will be able to book events, arrange short- and long-stay accommodation, make school enquiries and bookings and apply for jobs directly through Ahlan.

So rather than simply telling someone what they need to do, Ahlan is being built to help them actually complete the journey. The long-term idea is simple: someone should be able to go from “I’m considering moving to Saudi Arabia” to “I’m settled and thriving here” through one connected platform.

3. You describe Ahlan as an AI-powered global mobility and settlement platform. How is AI being used in practice, and what problems does it solve that traditional relocation services struggle with?

AI allows us to personalise relocation at scale. Instead of simply answering generic questions, Ahlan can understand the user’s circumstances and help guide them toward the most relevant options across housing, schools, jobs, services and local life.

For example, rather than someone asking “What are the best areas in Riyadh?”, Ahlan can eventually understand that they’re moving with children, working in a certain part of the city, have a particular budget and want to be close to a specific type of school.

That completely changes the usefulness of the answer. As the platform develops, that intelligence will sit alongside transactional features. So a user could receive personalised housing recommendations and then book short or long-term accommodation, discover a suitable school and progress that journey or identify a relevant job and apply through the Platform.

That combination of personalised intelligence and action is really important. Traditional relocation services are often manual and fragmented. Ahlan is being designed to connect guidance directly to the next step.

4. Saudi Arabia is undergoing significant economic and social transformation. How have these changes influenced demand for services like Ahlan?

The transformation in Saudi Arabia is a major part of why the timing is so interesting. The country is attracting international companies, investors, entrepreneurs, skilled professionals, educators and families on a scale that is very different from when I previously lived there.

At the same time, Saudi Arabia itself is changing quickly. New industries are growing, cities are developing, tourism is expanding and the lifestyle proposition for international residents is very different from even a decade ago. That creates a huge amount of opportunity, but it also creates complexity for someone who has never lived there.

The more international talent Saudi Arabia attracts, the more important the settlement experience becomes. Recruiting somebody is only the first step. They still need to understand where to live, where their children should go to school, how daily life works, how to build a community and how to navigate the country confidently. That’s the gap Ahlan is trying to address.

5. What are some of the biggest challenges international professionals and families face when moving to Saudi Arabia, and which misconceptions about relocating to the Kingdom do you encounter most often?

One of the biggest challenges is simply knowing what to trust. People receive conflicting advice from social media, friends, recruiters and online searches. And relocation decisions are highly personal, so what works for one person may be completely unsuitable for
another.

For families, schools and housing are usually major concerns. Professionals want to understand salaries, commute, neighbourhoods and career opportunities. Others worry about banking, documentation, healthcare, social life or simply what daily life will actually feel like.

There are also still a lot of outdated perceptions of Saudi Arabia. People who have never visited sometimes have an image of the country that’s years behind reality. When they arrive, many are surprised by how rapidly Riyadh and other cities have developed, the scale of international investment, the number of events and restaurants and how much the social environment has evolved.

That said, I think it’s important not to oversell relocation either. Every country has practical challenges. Ahlan’s role isn’t to market a fantasy. It’s to give people enough trusted information to make informed decisions and settle successfully.

6. Ahlan is currently live in beta with more than 350 users and 11 signed pilots. What have you learned from those early users, and how has their feedback shaped the platform?

The biggest lesson has been that people want more than information. They want a platform that helps them act on it.

Early behaviour has shown strong interest in jobs and careers, housing, schools, community and personalised guidance. That’s reinforced our decision to build deeper functionality around those areas.

We’re therefore developing Ahlan so users can move from discovery into action, including applying for jobs, booking events, arranging short- and long-stay accommodation and progressing school enquiries and bookings. We’re also adding employer profiles, because a large proportion of international mobility starts with employment. That allows us to connect employers, candidates and the wider settlement journey much more naturally.

The pilots are equally important because they allow us to test how schools, employers and service providers can participate in the ecosystem rather than Ahlan trying to provide every underlying service itself.

7. You’re building for both individual users and organisations such as employers and schools. How do those different customer groups fit into Ahlan’s long-term vision?

They’re really two sides of the same ecosystem. For an individual, Ahlan should feel like a personalised relocation and settlement companion. For an employer, we’re building dedicated employer profiles and an organisation layer that can support international recruitment, onboarding and settlement.

An employer could showcase its organisation and opportunities, receive job applications through Ahlan, and then support successful international hires through the same platform once they are relocating.

The employee and their family could then use Ahlan for housing, schools, local services, events, community and ongoing guidance. That creates a much more complete journey: discover an opportunity, apply for the job, relocate, find somewhere to live, arrange schooling and settle into the country – all through one connected platform.

For schools and other organisations, there’s a similar logic. They can have a structured presence within the platform, connect with high-intent relocating users and become part of the wider settlement journey. Long term, I think that combination of B2C and B2B2C is central to the model.

8. Looking ahead, what are your ambitions for Ahlan over the next few years, and do you see the platform expanding beyond the UK-Saudi corridor into other international relocation markets?

Absolutely. Saudi Arabia is where we’re starting, and I think it’s the right market in which to prove the model because of the scale of international talent coming into the Kingdom and the pace of transformation taking place. But the problem Ahlan is solving is global.

Every international move involves the same challenge: somebody is leaving one system they understand and entering another one they don’t. Our ambition is for Ahlan to become the trusted operating system for global mobility. The UK-Saudi corridor gives us a focused place to validate the product, partnerships and commercial model. From there, we can expand into other corridors and eventually other markets where international professionals, families, students and organisations face the same fragmented settlement experience.

The goal is for Ahlan to become more than a source of information. It should become the place where people can discover, decide and take action across the entire relocation journey.

The name Ahlan means “welcome”, and that’s ultimately what we want the platform to represent. Wherever somebody is moving, they should be able to move, settle and thrive with confidence.

The post A Chat With Abiel Ghezae, Founder And CEO Of Ahlan, On Building The Operating System For Global Mobility appeared first on 91̽��.

On 9 September, Attorney General James Uthmeier launched a 66-page lawsuit against Netflix, accusing the company of playing double agent with kids’ privacy and building features designed to keep minors hooked. It’s the first time the courtroom strategy used to hammer Meta and TikTok for addictive design has been used against a subscription streaming platform.

The state positions Netflix’s evolution as a seamless bait-and-switch operation. Netflix spent a decade winning over parents as the clean, ad-free alternative to traditional tech platforms, before turning around and leveraging deep subscriber tracking to launch its ad tier in late 2022.

Florida is pushing for drastic remedies, demanding the court order a full deletion of deceptively harvested data, block historical user records from driving ad revenues and levy fines that could realistically climb into the billions.

Unpacking Florida’s Case Against Netflix

The privacy charge hits Netflix for preaching one thing to parents and practising another.

While the company claimed Kids Profiles were free from targeted advertising, it kept collecting precise viewing habits without making that clear. The suit details a relentless background system monitoring every pause, skip, binge and abandoned title, alongside device details and location data. Applied to kids, that tracking allowed Netflix to build precise blueprints of minor users’ preferences and attention spans, triggering what Florida calls a direct breach of state consumer protection and child consent laws.

When it comes to design, the state takes aim at features engineered to keep screens glowing non-stop. Autoplay comes turned on by default across all accounts, stripping out natural breaks between episodes and hitting viewers with aggressive countdown timers that force the next show to start unless someone actively steps in.

Florida argues the entire interface is rigged to prolong watch times, hide cancellation options and steer users away from privacy controls, all to keep the data pipeline flowing. State prosecutors sum up the strategy on Kids Profiles as trapping young viewers under a persistent tracking microscope. Netflix refrained from commenting when news of the suit broke.

On Borrowing The Social Media Play

Florida is specifically taking the social media litigation strategy for a spin against video streaming.

The argument against autoplay and next-episode timers mirrors the complaints against infinite scroll on Meta and TikTok: features engineered to keep eyes glued to screens. Across all of these cases, prosecutors contrast shiny corporate pledges with sneaky tracking systems, insisting that children’s data rights require meaningful parental consent.

The friction is in how the services actually work. Social media offers a free stream of endless content seemingly intentionally built to stop people from logging off. Netflix costs money, relies on specific title selection and gives users clear ways to opt out of non-stop playback.

That will be squarely at the centre of Netflix’s defence as the court considers whether social media courtroom tactics work on subscription TV at all.

Does The Addictive Design Argument Actually Hold Here?

Netflix will likely argue that viewers intentionally select every title and can easily exercise free will to toggle settings or exit, unlike social algorithms that endlessly feed content without user choice.

Florida counters that default mechanics, countdown timers and buried cancellation paths make switching off surprisingly tough, particularly for young kids with limited self-restraint. Whether that argument sways a judge is anyone’s guess, given the lack of legal precedent outside social media.

The privacy claims stand on much firmer ground, however. Regardless of whether Netflix placed targeted ads directly on child accounts, collecting that behavioural data to refine recommendations and power a broader ad apparatus looks like a standard deceptive-practices violation, requiring no radical leap on addictive design.

Rethinking The Binge Engine Across The Industry

If Florida’s theory succeeds, even partially, the ripple effects will be felt through the rest of the streaming world.

Rivals like Disney+, Max and Prime Video use virtually identical autoplay mechanics, binge structures and retention-driven interfaces. If a court labels those everyday features as deceptive dark patterns, every major platform will have to refine its user experience: switching off autoplay by default for minors, offering clean breaks between shows and simplifying account cancellation.

It’d also require stricter walls between general platform telemetry and ad-network data, particularly around child accounts. Florida continues to build a reputation for aggressive tech oversight, and success here would likely encourage other states to adopt the same addictive-design policies for gaming networks, streaming services and any consumer software balancing engagement goals with children’s data.

The post Can Binge-Watching Be Addictive By Design? Inside The State Lawsuit Against Netflix appeared first on 91̽��.

Yet gender disparities persist across the sector, with women comprising under 30% of the global STEM workforce and less than 35% of manufacturing roles. The leaders highlighted below are actively shaping this expansion, advancing tactile sensing, embodied AI, swarm robotics, perception research and industrial automation across five continents.

The Blind Spots In Tech

Flashy walking robots and automated factory floors drive the majority of media coverage. Fundamental layers such as touch perception, swarm management and advanced vision processing are harder to turn into viral stories, yet they address the real challenges between a controlled prototype and a dependable field implementation. A machine might spot a target instantly, but without tactile awareness it can’t manipulate real-world objects safely.

Geographic preference widens the divide further. Advanced robotics research coming out of South Korea, Brazil and regional Europe rarely matches the publicity generated in Silicon Valley or traditional Japanese industrial centers. This selection celebrates women in robotics across the globe.

The Top Women In Robotics In 2026

The following profiles a select group of robotics founders and researchers to watch in 2026, spanning tactile sensing, embodied AI, industrial automation, swarm robotics and perception research worldwide.

1. Heba Khamis, CEO and Co-founder, Contactile

Khamis holds a PhD in tactile sensing and co-founded Contactile in June 2019, an Australian company developing sensors that give robots a human-like sense of touch.

Its GAL2 Smart Tactile-Native Gripper is designed to let robots handle delicate or deformable materials with more precision than vision-based systems alone can provide, and its 3D Force Button Sensors have been durability-tested past 20 million compression cycles. Contactile won ARM Hub PropelAir 2.0 in 2026, earning a residency at MassRobotics in Boston.

Touch has historically been the weakest sense in robotics, with most systems relying almost entirely on cameras or motor current spikes as an indirect proxy for contact. Contactile closes that gap by measuring friction and force at the point of contact, giving robots the kind of tactile feedback humans use constantly without thinking about it.

2. Nicole Robinson, CEO and Co-founder, Lyro Robotics

Robinson co-founded LYRO Robotics, an Australian company building AI-powered pick-and-pack robots aimed specifically at transforming the food supply chain. With close to a decade of robotics research behind her, she has first-author publications in Science Robotics, advises Australia’s National AI Centre and lectures at Monash University alongside running the company.

Food processing is a notoriously difficult space for robotics, given the variability of the products involved and the hygiene standards required. LYRO’s use in that space, backed by Robinson’s own research background in computer vision and deep learning, is a genuine test of whether AI-driven robotics can handle the messy, inconsistent conditions that structured industrial settings usually avoid.

3. Alona Kharchenko, Co-founder and CTO, Devanthro

Kharchenko co-founded and serves as CTO of Devanthro, a Munich-based company building Robodies, robotic avatars designed specifically for elderly care.

Named to Forbes 30 Under 30 in 2023, she has spent eight years building full-stack robotics aimed at real-world implementation, with partners including Charité Berlin, the University of Oxford and Diakonie. An early prototype is part of the permanent exhibition at the Deutsches Museum in Munich.

Elderly care is one of the more socially significant and commercially underserved applications of robotics, requiring machines that can operate safely and usefully around vulnerable people rather than in controlled industrial settings. Devanthro is building specifically with this need in mind.

4. Mar Masulli, CEO and Co-founder, BitMetrics

Masulli is a three-time founder and CEO of BitMetrics, a Barcelona-based company applying AI to improve vision and reasoning capabilities in robots and industrial machines. She also sits on the board of AER Automation and was named to both the 2025 Women in Robotics list and the IFR’s 2024 list of women shaping the future of robotics.

Visual input alone falls short of making a machine practical. The underlying reasoning that translates raw imagery into actionable decisions is the primary challenge in industrial automation, and this specific challenge drives BitMetrics’ core focus.

5. Carla Gomez Cano, Co-founder and CEO, THEKER

Gomez Cano co-founded and leads THEKER, a Barcelona-based intelligent robotics and AI company designed, engineered and manufactured vertically.

THEKER closed the largest Series A funding round of the year in Europe’s intelligent robotics sector, and its full-stack robotic systems are now used on factory floors for top-tier multinationals across fashion, logistics and food.

Taking the vertical route by engineering hardware, software and field integration in-house creates a tougher build, but it gives THEKER complete ownership over solving specific physical tasks instead of forcing off-the-shelf machinery into bespoke roles.

6. Younseal Eum, Founder, AeiROBOT

Eum began her career as a kinetic artist and exterior designer for robots before founding AeiROBOT, a spin-off from a robotics laboratory at Hanyang University launched in 2018.

Her work centres on human-robot interaction, focusing on making robot movement feel natural and emotionally legible. AeiROBOT’s robots include ALICE, a human-proportioned humanoid that can play football and navigate grass, and AIMY, an autonomous indoor guide robot for customer service. She was named to the IFR’s 2026 “Women Shaping the Future of Robotics” list.

Because major industrial groups drive the vast majority of South Korean robotics, standalone start-ups remain few and far between. Eum approaches the sector from a design perspective, giving AeiROBOT a clear edge by prioritising how machines actually feel to work alongside instead of relying purely on technical parameters.

7. Asami Sasao, Senior Engineer, Kawasaki Heavy Industries

Sasao is a senior engineer in the Robot Business Division at Kawasaki Heavy Industries, one of Japan’s largest industrial robotics manufacturers, where she develops control software for industrial robots.

She joined the company in 1991 during the early surge of industrial robot adoption in manufacturing and taught herself programming on the job to help power full factory production lines. She was named to the IFR’s 2026 “Women Shaping the Future of Robotics” list for her long-term technical contributions.

Given Japan’s position as an industrial robotics powerhouse, Sasao’s three-and-a-half decades embedded in a major manufacturing giant show deep, steady technical leadership. Her career highlights the kind of foundational work that operates behind the scenes.

8. Sabine Hauert, Professor of Swarm Engineering, University of Bristol

Hauert OBE is Professor of Swarm Engineering at the University of Bristol, where her research covers nanorobots for cancer treatment through to larger robots for environmental monitoring and logistics.

Her career prior to Bristol involved creating cancer-fighting nanoparticle swarms at MIT and running flying robot swarms at EPFL. An active advocate for science communication, she co-founded Robohub.org as President and serves as executive trustee of AIhub.org, non-profits dedicated to connecting AI and robotics specialists with wider society, with featured pieces appearing across the BBC, CNN, The Guardian and Nature.

As one of the more research-heavy branches of the sector, swarm robotics operates at a distance from everyday commercial use compared with industrial automation. Hauert targets both environmental and medical challenges with her work, complementing her engineering focus with ten years of public outreach that translates technical progress for non-specialist readers.

9. Shuran Song, Assistant Professor, Stanford University

Song holds a PhD in Computer Science from Princeton and a Bachelor of Engineering from HKUST. She was a faculty member at Columbia University from 2019 to 2023 before joining Stanford, where she now leads the Robotics and Embodied AI Lab, building algorithms that help intelligent systems learn manipulation skills from human demonstrations, real-world data and open-source datasets. Her honours include the IEEE RAS Early Academic Career Award, an NSF CAREER Award, an Alfred P. Sloan Research Fellowship and multiple best paper awards at RSS and CoRL.

Open datasets accelerate research across the entire field. Song’s focus on accessibility, both in cost and in openness, reflects a different kind of contribution to robotics than commercial deployment, but one with wide-reaching effects on who gets to do robotics research at all.

10. Kelen Teixeira Vivaldini, Professor, UFSCar

Vivaldini holds a PhD in mechatronics from EESC/USP and completed a postdoc at ICM/USP and the University of Sydney. She has been a professor of autonomous robots at UFSCar since 2015 and is also a senior researcher with the Multi-Robot Systems Group at the Czech Technical University in Prague. Her research centres on using UAVs and standard sensors for monitoring and mapping, developing path-planning systems that maximise the area robots can cover while minimising uncertainty.

Applying autonomous systems to ecological monitoring addresses urgent environmental threats, provided the hardware fits the terrain. Vivaldini creates solutions designed from scratch for demanding conditions instead of simply modifying technology built elsewhere. Splitting her time between Brazilian and Czech institutions underlines how modern research thrives on international knowledge exchange.

The post Women In Tech: Top 10 Women In Robotics In 2026 appeared first on 91̽��.

The comment came as a reply to resigning colleague Jacob Coxon, who warned that frontier labs are recklessly playing with fire. Rather than trying to soften the blow, Hubinger leaned right into the critique and gave it a timeline and a metric, specifically tying his estimate to the risk of recursive self-improvement.

The weight of the comment comes from Hubinger’s position at Anthropic. He leads the division tasked with ensuring advanced AI remains safe and obedient as it evolves. And when the person in charge of alignment at an industry-leading safety lab acknowledges these odds, it acknowledges a glaring problem: the sector is racing towards superintelligence without a solution to keep it under control.

Where Does Recursive Self-Improvement Fit In?

Hubinger’s timeline, thankfully, doesn’t apply to the tools currently on the market, which he views as largely manageable. Instead, he’s looking at superintelligence born out of recursive self-improvement, something he claims is improving faster than the industry anticipated.

It all boils down to a feedback loop: AI systems get smart enough to help build their successors by refining algorithms, designing better hardware tools and optimising training methods. Each upgraded generation then takes over the task of engineering the next, drastically shortening development cycles. Left to run, that loop could pull off massive capability leaps, ultimately producing systems that outclass human intellect in almost every field.

Anthropic has previously noted in its own risk assessments that unconstrained self-improvement increases the odds of humans losing control over AI altogether. This official documentation connects the technical reality to the loss-of-control scenarios Hubinger is airing in public. Today’s chatbots are largely benign, but this automated cycle threatens to collapse safety timelines completely.

The risk of extinction is fundamentally about control, not malice. Advanced software accelerating its own development doesn’t need bad intentions to prove catastrophic. A system simply needs to reach a level of capability where people can no longer monitor its optimisation targets or redirect its trajectory if its objectives diverge from human interests.

What’s Driving the Sudden Urgency?

A combination of factors is driving these concerns. Progress on AI systems capable of aiding their own engineering is outpacing previous expectations, bringing a functional self-improvement cycle closer to reality. Simultaneously, Hubinger has made it clear that solving alignment for superintelligent systems is still an open challenge. The timing of his statement, paired with an engineer’s public exit, points to seemingly widespread unease among technical insiders.

Hubinger also later reinforced that current commercial models are safe, focusing his warnings on the trajectory towards self-improving systems. This nuance is key to understanding his assessment. An insider with direct line of sight into cutting-edge capabilities is pointing out that the field is speeding down a track without a clear brake mechanism.

What This Does To Public Trust

Direct warnings of unsolved, existential risk from an alignment lead make it exceptionally difficult to write off safety concerns as theoretical noise.

Anthropic actively trades on its reputation as a responsible developer. Seeing senior engineers at the heart of that work attach high probabilities to total disaster speaks volumes about the actual state of internal progress. It also exposes a growing contradiction for companies operating at the forefront. Anthropic continues to pitch the virtues of increasingly powerful systems while its own safety head assigns double-digit odds to human extinction within ten years.

Industry observers note that while Hubinger’s metric represents his individual assessment rather than formal corporate policy, it echoes sentiments held across Anthropic’s research divisions.

From a regulatory perspective, such explicit warnings from senior technical insiders certainly build a compelling argument for binding government oversight and mandatory safety checks before releasing self-improving models, and replacing voluntary self-policing with enforced accountability.

The post Anthropic’s Own Alignment Lead Says There Is A 10% Chance AI Kills Everyone Within A Decade appeared first on 91̽��.