Signal has accused the UK government of proposing a dangerous form of surveillance after ministers announced plans that could require technology companies to prevent children from taking, sharing, or viewing nude images on smartphones and tablets.

What Is The Government Proposing?

The announcement came from Prime Minister Keir Starmer during London Tech Week, where he said the UK would become “the first country in the world to make it impossible for children to take, share or view nude images.”

Under the proposals, technology companies including Apple and Google would be expected to activate existing safety features or introduce new technical measures that detect and block nude images on devices used by children. Adults would still be able to access such content after completing age verification checks.

Three-Month Deadline

The government has given technology companies three months to develop suitable solutions. If they do not, ministers have indicated they are prepared to introduce legislation, financial penalties, and potentially other enforcement measures.

The government argues that stronger intervention is needed because online child sexual abuse, exploitation, and exposure to harmful content remain widespread. Home Office figures cited alongside the announcement indicate that 91 per cent of online child sexual abuse reports recorded in 2024 contained self-generated content from children themselves.

Why Is Signal Opposing The Plan?

Signal, one of the world’s best-known encrypted messaging platforms, has responded forcefully to the proposals. In a public statement, the company said the government’s approach “will not safeguard children. It endangers us all.”

The company’s main concern is not the goal of protecting children, but the technology required to achieve it. Signal argues that forcing devices to scan content before it is viewed, shared, or stored would create a new form of surveillance infrastructure capable of examining private information on users’ devices.

According to Signal, “Forcing all UK residents to prove their age and/or have all their content scanned, simply to exercise their fundamental right to communicate, is a perilous proposition.”

The company also warned that once such capabilities exist, they rarely remain limited to their original purpose. Signal stated: “We know that mass surveillance and censorship capabilities, however sincere-sounding the promises of those who initiate them are, never remain narrowly scoped.”

The Debate Around Client-Side Scanning

At the centre of the controversy is a technology known as client-side scanning.

Unlike traditional content monitoring, which takes place on external servers, client-side scanning operates directly on the user’s device. Supporters argue this provides a compromise between privacy and safety because images do not need to be sent elsewhere for inspection.

Advocates say the approach can prevent harmful content from being created, viewed, or shared while keeping personal information on the device itself.

Critics, however, argue that the distinction is not as clear-cut as it appears.

Although images may never leave the device, the device is still examining content on behalf of a third party. Privacy groups have long argued that this changes the fundamental trust relationship between users and their devices.

Signal’s concern is that the same scanning infrastructure could potentially be expanded in future to identify other forms of content beyond child protection material. Whether or not such powers were ever used, critics argue that the capability itself creates new risks around surveillance, censorship, security vulnerabilities, and public trust.

A Wider Privacy Battle

The disagreement reflects a much broader debate that has been developing for years.

Previous UK legislation, including the Investigatory Powers Act and aspects of the Online Safety Act, has generated similar disputes between governments seeking stronger online protections and privacy advocates concerned about the long-term consequences of expanding monitoring powers.

Technology companies have also faced these questions before. Apple, for example, previously proposed a system for detecting child sexual abuse material on devices before ultimately abandoning the project following widespread criticism from privacy and security experts.

Supporters of the government’s latest proposals argue that child protection must take priority. Organisations including the NSPCC, Internet Watch Foundation, Barnardo’s, and the Children’s Commissioner for England have publicly welcomed the plans.

NSPCC chief executive Chris Sherwood described the proposal as “a major step forward in our fight against online child sexual abuse.”

What Does This Mean For Your Business?

The wider significance of this dispute isn’t really about nude image detection. It is about where governments, technology companies, and citizens draw the line between child protection and personal privacy, particularly when proposals involve technology capable of examining content directly on people’s devices.

The debate also highlights a growing tension that businesses are increasingly encountering across cyber security, compliance, artificial intelligence, and digital regulation. Governments are seeking stronger protections against genuine harms, while technology providers and privacy advocates are warning about the unintended consequences of expanding monitoring capabilities.

The larger issue here is not simply whether children should be protected online, as few would disagree with that objective. The real debate is whether it is possible to achieve those protections without creating technologies that examine private content on personal devices. As governments around the world continue to grapple with that question, the outcome is likely to influence the future of privacy, encryption, and digital communications far beyond the UK.

A small but growing group of businesses is now spending thousands of dollars per employee every month on artificial intelligence, suggesting that AI is increasingly being treated as core business infrastructure rather than simply another productivity tool.

The Rise Of The “AI-Pilled” Company

The findings come from the latest Ramp AI Index, which analyses anonymised spending data from more than 70,000 US businesses to track how organisations are adopting AI.

Until recently, most AI adoption studies focused on whether businesses were using AI or not. Ramp now believes that question is becoming less useful as AI adoption becomes increasingly widespread. Instead, the company is focusing on what it calls the “intensity of adoption”, i.e., how heavily businesses are actually investing in AI.

One of the report’s most eye-catching findings is that the top 1 per cent of firms, described by Ramp as “AI-pilled”, are spending an average of $7,449 per employee per month on AI services. By comparison, the top 10 per cent spend around $611 per employee, while the median business spends just $11.38, roughly equivalent to a single ChatGPT or Claude subscription.

The figures highlight just how wide the gap is becoming between businesses experimenting with AI and those building it deeply into everyday operations.

What Does “AI-Pilled” Mean?

The tech sector term “AI-pilled” essentially describes organisations that have moved beyond occasional AI use and started treating AI as a core operating model.

At Ramp itself, chief product officer Geoff Charles recently outlined how the company achieved 99.5 per cent AI adoption among employees, with more than 1,500 internal applications reportedly created in six weeks by over 800 different staff members.

The goal is not simply to give employees access to chatbots. Instead, it involves embedding AI into workflows, automating routine tasks, building internal tools, deploying coding agents, and allowing staff across multiple departments to use AI as part of their daily work.

In these organisations, AI is increasingly viewed as a business capability rather than a software product.

Still Cheaper Than Hiring People

Despite the impressive spending figures, Ramp’s research suggests that AI has not yet reached the point where organisations are routinely spending more on AI than on employees.

The report notes that “the top 1 per cent of firms spend $7.45K per employee per month” but also points out that this remains less than half the typical monthly salary of a software engineer.

That finding is important because it challenges some of the more dramatic claims surrounding AI adoption.

Recent headlines have highlighted companies spending heavily on AI agents, tokens, and computing power, while some technology executives have suggested AI could eventually become a larger cost centre than human labour.

For now, however, the data suggests that even the most advanced adopters continue to view AI as something that augments employees rather than replaces them entirely.

Why Spending Continues To Rise

Perhaps the most significant finding is not how much these firms are spending, but how quickly that spending is growing. For example, Ramp found that the top 1 per cent of AI users increased spending per employee by 14.1 per cent in a single month.

This is happening despite growing awareness of AI costs and increasing efforts to use cheaper models where possible.

The report notes that many businesses are actively seeking lower-cost alternatives, including open-source models and newer competitors such as DeepSeek, yet overall spending continues to climb.

One explanation is that businesses are moving from occasional AI usage towards much broader deployment. As organisations connect AI into customer service, software development, analytics, administration, marketing, finance, and operations, overall consumption naturally increases.

The more AI becomes embedded into business processes, the more computing power, tokens, APIs, and specialist tools are required to support it.

No Single Vendor Dominates

Another interesting finding is that the most advanced AI users are not putting all their eggs in one basket.

According to Ramp, “advanced usage of AI means using multiple frontier models”, alongside platforms providing access to open-source models and specialist AI-native software.

This suggests that businesses are becoming increasingly sophisticated in how they approach AI procurement.

Rather than committing exclusively to one provider, many organisations appear to be selecting different models for different tasks based on performance, capabilities, security requirements, and cost.

That approach mirrors earlier developments in cloud computing, where organisations often adopted multi-cloud strategies to reduce dependence on a single supplier.

Why This Matters

The wider significance of Ramp’s findings is not really about the $7,500 figure. The more important story here is that a growing number of businesses now appear to view AI as an operational resource that sits alongside software, cloud infrastructure, and human expertise.

For years, technology adoption was largely measured by whether organisations used a tool at all. Increasingly, the competitive gap may depend on how deeply AI becomes integrated into workflows and decision-making processes.

The data also suggests that AI adoption is becoming far more uneven. While many businesses remain at the subscription stage, a small group of early adopters is investing heavily and experimenting with entirely new ways of operating.

What Does This Mean For Your Business?

For businesses, the report highlights an important distinction between using AI and building around AI.

Most organisations are unlikely to spend thousands of dollars per employee each month on AI in the foreseeable future. However, the research suggests that some companies are already treating AI as a strategic capability worthy of significant ongoing investment.

That doesn’t necessarily mean every business should increase its AI budget dramatically. Ramp’s figures measure spending, not outcomes, and high expenditure alone does not guarantee productivity gains or return on investment.

The more useful lesson here may be that leading adopters are moving beyond standalone chatbots and experimenting with AI agents, automation, workflow integration, and multi-model strategies. As the technology continues to mature, the organisations that learn how to apply AI effectively across their operations may gain a greater advantage than those that simply spend the most money on it.

A German court has ruled that Google can be held directly responsible for false information generated by its AI Overviews feature, a decision that could have significant implications for AI-powered search engines and chatbots worldwide.

What Happened?

The case centres on Google’s AI Overviews, the AI-generated summaries that increasingly appear at the top of search results and attempt to answer users’ questions directly without requiring them to visit other websites.

The ruling came after two German publishers discovered that AI Overviews had falsely associated them with scams, subscription traps, and questionable business practices. According to court documents, the AI-generated summaries created links and allegations that did not appear in the sources cited by Google.

The Regional Court of Munich issued an injunction preventing Google from repeating the statements and concluded that the AI-generated content should be treated as Google’s own speech rather than merely a summary of information published elsewhere.

Why The Court Reached This Decision

The most significant aspect of the ruling is the distinction the judges drew between traditional search results and AI-generated answers.

For many years, search engines have generally benefited from limited liability protections because they primarily act as intermediaries, directing users to content created by third parties. If a search result links to an inaccurate webpage, responsibility normally rests with the publisher of that page rather than the search engine itself.

AI Overviews operate differently. According to the Munich court’s judgment, AI Overviews do not simply display links or snippets. Instead, they create “independent, new and substantive statements” that are generated by Google’s AI systems and presented to users as complete answers. The judges noted that Google controls the AI model and the algorithms that produce these summaries and therefore has responsibility for the content they generate.

The court also highlighted that the disputed statements were not simply copied from third-party websites. In several cases, the AI generated connections and allegations that were not present in the source material at all.

Why Google’s Defence Failed

Google argued that users can inspect the sources linked within AI Overviews and therefore verify information for themselves. However, the court rejected that argument.

The judges compared AI Overviews to headlines or teaser text that many readers consume without investigating further. They concluded that if an AI-generated answer is presented as a complete and understandable response, the fact that users could conduct additional research does not remove responsibility from the party that published it.

The court also found that AI Overviews are not essential to the functioning of search in the same way that traditional search results are. Users can still find information through ordinary links without requiring an AI-generated summary.

That distinction helped the judges justify imposing a higher level of responsibility on Google for AI-generated content than has historically applied to search engines.

Google Plans To Appeal

Google has confirmed that it intends to challenge the ruling. A company spokesperson told Reuters that the case focuses on specific errors rather than the fundamental operation of AI Overviews and said the company disagrees with the court’s conclusions.

Google also argued that the overwhelming majority of AI Overviews are accurate, while acknowledging that AI systems can occasionally miss context or misinterpret information. The company says it takes action when policy violations are identified.

The case remains a preliminary injunction rather than a final appellate ruling, meaning the legal position could still change as appeals progress.

A New Liability Challenge For AI

AI systems increasingly generate answers by analysing information from multiple sources and presenting users with a single, consolidated response. That approach is now common across search engines, chatbots, virtual assistants, and business productivity tools.

The Munich court’s reasoning suggests that once an AI system begins creating its own narrative, combining information from multiple sources and generating new conclusions, the provider may no longer be able to rely on the legal protections traditionally available to search engines and hosting platforms.

If that principle survives appeal, it could affect a wide range of AI products, including AI-powered search tools, enterprise assistants, customer service bots, and generative AI platforms.

Also, regulators across Europe are paying closer attention to AI transparency, accountability, and safety. Questions about who should be responsible when an AI system produces false, defamatory, or harmful information are becoming increasingly important as these tools become more widely used by businesses and consumers alike.

What Does This Mean For Your Business?

Most businesses are unlikely to see any immediate operational impact from the ruling, but it highlights an issue that organisations should already be considering.

AI-generated answers can appear authoritative and convincing while still containing errors, misunderstandings, or entirely fabricated information. As AI tools become more deeply embedded into search engines, productivity software, and business workflows, organisations should avoid treating AI-generated content as automatically accurate.

The ruling also serves as a reminder that businesses should monitor how AI systems describe their brands, products, and services online. If an AI-generated summary contains false information, legal avenues for challenging those statements may become clearer if courts increasingly view AI output as the responsibility of the platform that generated it.

More broadly, the case signals that regulators and courts are beginning to move beyond the question of what AI can do and focus instead on who should be accountable when it gets things wrong.

It’s been reported that Meta has quietly removed facial recognition code from the companion app used by its AI-powered smart glasses, reigniting concerns about how far wearable technology companies may be willing to go in their pursuit of always-on artificial intelligence.

What Was Removed?

The controversy centres on an internal system called NameTag, which was discovered inside the Meta AI smartphone app that works alongside the company’s Ray-Ban smart glasses. According to reporting first published by WIRED, the code appeared to support facial recognition capabilities that had never been publicly released.

The system was reportedly designed to convert faces captured by the glasses into unique biometric identifiers, often referred to as faceprints, and compare them against a database stored on the user’s device. Evidence within the software also suggested that faces the system could not identify would be cropped, indexed, and stored locally for future processing.

Most notably, the code was present inside an application installed on tens of millions of devices despite Meta repeatedly stating that no final decision had been made about introducing facial recognition to its smart glasses platform.

Just one day after the findings became public, Meta released an updated version of the app that removed almost all traces of the NameTag system.

Meta’s Response

Meta says the facial recognition system was an internal exploratory project rather than a planned product feature. However, the speed with which the code was removed has inevitably attracted attention.

Reports indicate that the original software contained multiple AI models dedicated to detecting faces, cropping facial images, and converting them into biometric signatures. The app also reportedly contained a “Person recognised” alert that would have been displayed if someone was successfully identified.

Meta has not publicly explained why the code was removed immediately after the discovery or whether the changes had already been planned before the reporting appeared.

Why Facial Recognition In Glasses Is Different

The debate is not really about facial recognition itself. The technology has existed for many years and is already widely used in smartphones, airports, security systems, and consumer applications.

What makes smart glasses different is that they allow facial recognition to move from fixed locations and deliberate actions into everyday social interactions.

Unlike a phone, which requires someone to consciously point a camera at another person, smart glasses can continuously capture information while being worn. Combined with AI, cameras, microphones, and internet connectivity, they create the possibility of real-time identification in public spaces without the knowledge of the people being observed.

Supporters argue that such technology could have legitimate uses. For example, facial recognition could help visually impaired users identify friends, family members, or colleagues. It could also assist people with memory difficulties or cognitive impairments.

Critics, however, have raised concerns that the same technology could be misused for stalking, harassment, surveillance, or the identification of strangers without consent.

Those concerns become even more significant when combined with generative AI systems capable of searching, analysing, and contextualising information automatically.

Part Of A Bigger Strategy

The discovery also provides an insight into Meta’s longer-term ambitions for wearable AI. For example, chief executive Mark Zuckerberg has repeatedly described smart glasses as a future computing platform where AI assistants become constantly available throughout the day. The company’s recent investments in Ray-Ban and Oakley smart glasses reflect a belief that future digital interactions will increasingly move away from smartphones and towards wearable devices.

Facial recognition could potentially play an important role in that vision. An AI assistant capable of recognising people, understanding context, remembering previous interactions, and providing relevant information could become far more useful than one that simply responds to voice commands.

However, it is precisely that capability which raises difficult questions about privacy, consent, and personal data.

The Wider Privacy Challenge

The incident arrives at a time when regulators in Europe, the UK, and the United States are paying closer attention to biometric technologies.

Unlike passwords or usernames, biometric identifiers are linked directly to an individual’s physical characteristics. If compromised or misused, they cannot simply be changed or reset.

Privacy campaigners have long argued that facial recognition requires stronger safeguards than many other forms of personal data because of its potential to identify individuals at scale and without their active participation.

The rapid removal of the NameTag code suggests that Meta recognises the sensitivity of the issue, even if the company insists the feature was only exploratory.

What Does This Mean For Your Business?

For businesses, the story highlights how quickly AI is beginning to move beyond software and into the physical world.

Many organisations are already evaluating AI tools for productivity, automation, and customer service. The next wave of AI innovation is likely to involve wearable devices that can see, hear, interpret, and respond to the environment around them in real time.

That creates new opportunities, particularly in areas such as accessibility, training, field services, logistics, and hands-free information access. At the same time, it introduces new questions around privacy, data governance, consent, and the collection of biometric information.

The wider lesson is that as AI becomes more deeply embedded into everyday devices, businesses will need to think not only about what these systems can do, but also about what employees, customers, and the public are comfortable allowing them to do. The reaction to Meta’s facial recognition experiment suggests those conversations are only just beginning.

Anthropic has released a public version of the same AI technology that it previously restricted because of concerns about its cyber security capabilities, only for access to be suspended days after an intervention by the US government.

What Is Claude Fable 5?

Claude Fable 5 is a public version of Anthropic’s Mythos-class AI, a highly capable model originally developed for cyber security and vulnerability discovery work.

According to Anthropic, Claude Fable 5 and Claude Mythos 5 are “the same underlying model”, with the main difference being that Fable 5 includes additional safeguards designed to prevent misuse in areas such as cyber security, biology, chemistry, and model extraction. Mythos 5, by contrast, has some of those restrictions removed for approved users.

Anthropic originally developed Mythos-class models as part of Project Glasswing, a programme aimed at helping cyber defenders and critical infrastructure providers identify serious software vulnerabilities before attackers could exploit them.

When the first Mythos model was launched in April, Anthropic limited access to a small group of carefully vetted organisations because it believed the system’s cyber capabilities presented significant risks if made widely available.

Those concerns were not entirely theoretical. According to Anthropic, organisations using Mythos-class models have already identified “more than ten thousand high- or critical-severity vulnerabilities across the most systemically important software in the world”.

Why Anthropic Decided To Release It

Anthropic says it spent several months developing safeguards that would allow Mythos-level capabilities to be released more broadly while reducing the risk of misuse.

The result was Claude Fable 5, which the company described as “a Mythos-class model that we’ve made safe for general use”.

According to Anthropic, Claude Fable 5 delivers capabilities that were previously available only to a small group of approved organisations using Mythos. The company said: “Fable 5’s capabilities exceed those of any model we’ve ever made generally available.”

The model reportedly demonstrates state-of-the-art performance across software engineering, scientific research, vision tasks, analytical reasoning, and long-running autonomous work. Anthropic said it can “work autonomously for longer than any previous Claude models”, enabling it to tackle more complex tasks with less human supervision.

To reduce the risks associated with releasing such a powerful model, Anthropic introduced new safety systems that automatically redirect certain high-risk requests to a less capable model, Claude Opus 4.8.

According to the company, those safeguards were deliberately configured conservatively because “releasing a model this capable comes with risks”.

Suspended

However, just days after launch, Anthropic announced that access to both Fable 5 and Mythos 5 was being suspended.

The company said the US government had issued an export-control directive requiring it to disable access for foreign nationals, whether inside or outside the United States. Anthropic stated that the government believed it had become aware of a method for bypassing, or “jailbreaking”, Fable 5’s safeguards. A jailbreak is a technique designed to trick an AI system into ignoring or circumventing its built-in restrictions.

Challenged By Anthropic

However, Anthropic strongly challenged the significance of the alleged vulnerability. The company said it had reviewed the reported technique and found that it was capable of identifying only “a small number of previously known, minor vulnerabilities”. It also argued that comparable results could already be achieved using other publicly available frontier AI models.

Anthropic further stated: “We disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people.”

The company warned that applying the same standard across the industry could effectively prevent the release of future frontier AI models.

A New Kind Of National Security Debate

The dispute highlights a broader change taking place in how governments are approaching advanced AI. For example, historically, software products were largely regulated after release if problems emerged. Frontier AI models are increasingly being treated differently because of concerns that they may create new risks in areas such as cyber security, biotechnology, critical infrastructure, defence, and intelligence.

Anthropic itself appears to recognise that reality, and the company has repeatedly argued that governments should have the ability to intervene when genuinely dangerous models emerge. However, it also insists that such decisions should be transparent and supported by clear technical evidence.

In its response to the suspension order, Anthropic stated that governments should be able to block unsafe deployments “as part of a statutory process that is transparent, fair, clear, and grounded in technical facts”.

The disagreement therefore appears to be less about whether oversight is needed and more about where the threshold for intervention should sit.

Why This Matters

The release and subsequent suspension of Fable 5 suggests that AI developers are now reaching capability levels where some models may be viewed as strategic assets rather than ordinary software products.

That raises some difficult questions for regulators, governments, technology companies, and investors alike. If advanced AI models can genuinely accelerate vulnerability discovery, scientific research, software development, and other high-value activities, restricting access could slow innovation. However, if those same capabilities can be misused, governments may feel increasing pressure to intervene.

Anthropic appears to believe that tension will become increasingly common as frontier AI systems become more capable. The company has argued that governments should have powers to intervene where genuine risks exist, while also warning that overly broad restrictions could hinder beneficial uses of the technology.

The dispute over Fable 5 therefore highlights a growing challenge facing policymakers: deciding when an AI model should be treated as a normal commercial product and when it should be treated as a potential national security concern.

What Does This Mean For Your Business?

For businesses, the story highlights how rapidly the AI landscape is evolving beyond questions of productivity and automation.

Many organisations are still deciding which AI tools to adopt, yet policymakers are already debating whether some frontier models should be treated as potential national security concerns. That represents a notable change in how AI is viewed by governments.

The wider lesson is that future AI adoption may be influenced not only by technological progress but also by regulation, export controls, safety requirements, and geopolitical considerations. As AI systems become more capable, businesses may find that access to certain models, features, or services depends as much on policy decisions as on technical innovation.

The dispute over Fable 5 may ultimately be remembered as an early example of a much larger challenge: how to make increasingly powerful AI systems broadly available while still managing the risks that come with them.

Aviva says fraudsters are increasingly using AI-generated evidence to support fake or exaggerated insurance claims, particularly in motor insurance.

The insurer says it detected more than 18,400 fraudulent claims during 2025, worth an estimated £233 million if paid out.

According to Aviva, fraudsters are using altered accident photos, fabricated documents, inflated repair costs, and exaggerated damage reports. The value of fraudulent motor claims rose by 39 per cent during the year.

Pete Ward, head of claims counter fraud at Aviva, said: “We’re seeing fraud become more sophisticated, from exaggerated claims to the use of AI-generated documents.”

Businesses should be aware that AI can now create highly convincing fake images and documents, making independent verification of evidence increasingly important when assessing claims, transactions, or other high-value requests.