A reported attempt by the UK government to secure continued access to Anthropic’s most advanced AI models has highlighted how dependent many countries have become on frontier AI systems developed and controlled overseas.

What Happened?

The story centres on Claude Fable 5 and Claude Mythos 5, two of Anthropic’s most capable AI models.

Earlier this month, the US Commerce Department reportedly instructed Anthropic to suspend access to both systems following concerns about a technique that could be used to identify software vulnerabilities. The move followed reports that government officials had been alerted to a potential jailbreak affecting the models.

The restrictions quickly became an international issue because Anthropic’s most advanced systems are used by organisations far beyond the United States.

UK Asked For Exemption

Reports indicate that the UK government subsequently sought continued access to the models. However, no exemption was granted and the restrictions remained in place, leaving British users affected alongside other international customers.

Why Were The Models Restricted?

The restrictions stem from a disagreement about the risks posed by advanced AI systems with strong cyber security capabilities.

According to reports, researchers demonstrated a way of prompting Fable 5 to identify software vulnerabilities within computer code. Concerns were raised that such capabilities could potentially be used to support cyber attacks as well as cyber defence.

Anthropic strongly disagrees with that assessment. The company says the technique exposed only a limited number of previously known vulnerabilities and argues that similar capabilities already exist in other leading AI systems. Anthropic has also warned that applying this standard across the industry could severely restrict the deployment of future frontier AI models.

The dispute reflects a broader challenge facing policymakers. The same AI systems that can help defenders find and fix vulnerabilities can also potentially be used by attackers to identify weaknesses more quickly.

Why The UK Became Involved

The incident has drawn attention to the UK’s reliance on foreign AI providers.

Many British organisations increasingly use frontier AI models for software development, cyber security, research, data analysis, and operational tasks. Access to those capabilities is largely controlled by a small number of US companies.

Reports suggest that organisations in sectors including finance, healthcare, research, and government were affected when Anthropic’s models became unavailable.

The situation has also raised wider national security questions.

UK AI minister Kanishka Narayan reportedly highlighted the growing importance of advanced AI systems in areas such as cyber security, drones, and defence technologies, arguing that access to frontier AI is increasingly becoming a strategic issue rather than simply a commercial one.

Cyber Security Industry Pushback

The restrictions have generated significant opposition from within the cyber security community, where many experts argue that advanced AI models are becoming increasingly important defensive tools. For example, more than 80 cyber security leaders and researchers have reportedly signed an open letter calling for the measures to be reversed, including senior figures from major cyber security firms and technology companies.

Their concern is that security teams are already using frontier AI systems to identify software vulnerabilities, analyse malware, generate detection rules, and accelerate security research. From their perspective, restricting access to powerful AI models may reduce the ability of defenders to find and fix weaknesses before attackers can exploit them.

Critics also argue that determined attackers are unlikely to be deterred by the restrictions, given the growing availability of alternative frontier models, open-source systems, and overseas providers. The debate therefore centres on whether limiting access to advanced AI genuinely improves security or simply changes who is able to use the technology and for what purpose.

The Growing Case For Sovereign AI

One of the most important consequences of the dispute may be renewed interest in sovereign AI.

The term refers to a country’s ability to develop, host, control, or guarantee access to strategically important AI capabilities without relying entirely on foreign providers.

The UK has already launched a £500 million Sovereign AI Fund and other initiatives designed to strengthen domestic AI capabilities. The Anthropic restrictions are likely to be viewed by supporters of those programmes as evidence that greater technological independence may be necessary.

Similar conversations are now taking place across Europe, Canada, India, and other regions concerned about becoming dependent on a small number of foreign AI suppliers.

Why This Matters

The significance of the story extends well beyond Anthropic. For decades, most organisations assumed that software purchased from commercial suppliers would remain available unless a provider discontinued a product or suffered an outage. Advanced AI may not follow the same pattern.

The Anthropic episode demonstrates that frontier AI systems can become entangled in national security concerns, export controls, geopolitical tensions, and government interventions. Access can potentially be affected by decisions taken far beyond the control of the organisations using them.

The incident also illustrates how rapidly AI is moving from being a productivity tool to becoming a strategic technology with implications for economic competitiveness, cyber security, and national resilience.

What Does This Mean For Your Business?

For businesses, the immediate issue is not whether they use Anthropic specifically, but whether they understand their dependence on external AI providers.

Many organisations are integrating AI into software development, customer service, cyber security, research, and business operations. The Anthropic restrictions highlight that access to those capabilities may not always be guaranteed.

The wider lesson is that AI resilience may become as important as AI adoption. Organisations may increasingly need to consider where their AI services come from, what alternatives exist, and how dependent critical processes have become on specific providers.

The dispute also highlights a broader reality. As AI systems become more capable and strategically important, decisions about access may increasingly be influenced by government policy, national security considerations, and international politics as much as by technological innovation itself.

A brain-computer interface developed by researchers at the University of California, Davis, has enabled a man with advanced ALS to communicate with remarkable accuracy, return to full-time employment, and use a computer independently for nearly two years, marking one of the most significant real-world demonstrations of the technology to date.

How The System Works

The breakthrough centres on Casey Harrell, a man living with amyotrophic lateral sclerosis (ALS), a progressive neurological condition that destroys motor neurons and can eventually leave people unable to speak or move.

In 2023, surgeons implanted four microelectrode arrays into the speech motor region of Harrell’s brain. The arrays record neural activity associated with attempted speech, which is then analysed by machine-learning software developed by the UC Davis team.

The system translates those neural signals into phonemes, the basic sounds that make up words, before converting them into complete sentences. The decoded text can then be displayed on screen or spoken aloud using a synthesised version of Harrell’s voice from before ALS affected his speech.

According to the research paper published in Nature Medicine, the system achieved more than 99 per cent word accuracy during formal testing using a vocabulary of 125,000 words. Over nearly two years of real-world use, Harrell communicated more than 183,000 sentences, totalling almost two million words.

Moving Beyond The Laboratory

What makes the achievement particularly significant is that the technology was used independently at home rather than under constant supervision from researchers.

Many previous brain-computer interface studies have demonstrated impressive results in controlled laboratory settings. However, practical day-to-day use has remained a major challenge.

The UC Davis team reported that Harrell used the system for more than 3,800 hours over a 19-month period and operated it without researchers being present. After initial setup by trained care partners, he was able to communicate, browse the internet, send messages, participate in video calls, and control a computer cursor using only neural signals.

The researchers described this as one of the key barriers to real-world adoption that the project has now overcome.

In the paper, they wrote that the results demonstrate “that intracortical BCIs have the potential to support independent use in the home, marking a critical step toward practical assistive technology for people with severe motor impairment.”

Helping Someone Return To Work

The technology’s impact extends beyond technical performance metrics.

Despite being paralysed and unable to speak naturally, Harrell has returned to full-time employment as an environmental advocate while using the system. Researchers reported that he used the brain-computer interface as his primary method of communication, preferring it to previous assistive technologies.

The study states that the system enabled him to maintain “full-time employment” while independently managing professional and personal communications.

Harrell also highlighted the personal benefits of the technology. Speaking through the brain-computer interface, he said: “It is a life that is more full of dynamic action and with friends and family, with colleagues, and it is something that allows me to communicate more in my natural way of communicating than any other technology that I have experienced.”

Why AI Is Central To The Breakthrough

Although brain implants often attract the headlines, the most important innovation may actually be the software.

The hardware used in the project is based on existing microelectrode technology. The major advance comes from the AI-powered decoding system developed by the UC Davis team.

Their software platform, known as BRAND, uses machine-learning algorithms to interpret complex neural signals in real time and convert them into meaningful language. Researchers continually refined the algorithms during the study to improve accuracy, stability, and ease of use.

The research paper notes that the latest transformer-based decoder achieved a state-of-the-art word accuracy rate of 99.2 per cent while requiring little or no daily recalibration.

Important Limitations Remain

Despite the encouraging results, it should be noted here that the technology remains in the experimental stage.

The study involved only a single participant, and researchers acknowledge that it is not yet known how widely the results will apply to other patients with ALS or different neurological conditions.

The system also still relies on external computers, wired connections, and trained carers to connect the equipment each day. Widespread clinical use would require further miniaturisation, regulatory approval, and substantial reductions in cost.

The researchers themselves note that “future work will be needed to evaluate wireless or fully implantable systems, minimise setup time and expand access to users with different clinical profiles.”

What Does This Mean For Your Business?

For most organisations, brain-computer interfaces may seem far removed from everyday business concerns. However, the study provides another example of how AI is increasingly moving beyond software applications and becoming integrated with healthcare, assistive technologies, and human-machine interaction.

The achievement also highlights the growing role of AI in solving complex real-world problems that extend well beyond productivity tools and chatbots. In this case, machine learning is helping restore communication, digital access, and employment opportunities for someone who would otherwise face severe limitations.

The technology remains years away from routine commercial deployment, but the results suggest that brain-computer interfaces are beginning to transition from research projects into practical assistive tools. If future studies can replicate these results at scale, they could significantly improve quality of life for people living with ALS, paralysis, and other severe neurological conditions.

An Earth observation satellite has successfully identified targets on its own while in orbit, without requiring human analysts on the ground, marking what is believed to be the first reported use of a vision-language AI model operating in space.

What Happened?

The milestone took place aboard YAM-9, a satellite operated by space infrastructure company Loft Orbital.

Traditionally, Earth observation satellites collect large volumes of imagery and sensor data, which are then transmitted to Earth for analysis by either human operators or machine-learning systems. In this case, however, the analysis happened directly on the satellite itself.

Using software developed by NASA’s Jet Propulsion Laboratory (JPL) and Google’s Gemma 3 vision-language model, the spacecraft was able to interpret natural-language instructions and identify relevant features within the imagery it was collecting. According to reports, researchers asked the system to locate things such as infrastructure around railway hubs and areas where human development meets the natural environment, and the satellite successfully identified them.

The demonstration is believed to be the first publicly reported example of a vision-language model operating autonomously in orbit.

How The Technology Works

The project combined several technologies that have become increasingly important in artificial intelligence.

Vision-language models differ from conventional image-recognition systems because they can understand both images and natural-language instructions. Rather than being trained to identify only specific objects, they can interpret broader requests expressed in everyday language.

On YAM-9, Google’s Gemma 3 model was integrated into a software platform called NAVI-Orbital, developed by NASA JPL. The system ran on an Nvidia Jetson Orin AGX processor carried onboard the satellite.

This allowed the satellite to analyse imagery while still in orbit rather than waiting for instructions from Earth.

Instead of downloading vast quantities of raw data and asking analysts to search through it later, the satellite could determine which information was relevant and prioritise it automatically.

Why This Matters

The development could significantly change the economics and usefulness of Earth observation.

Modern satellites generate enormous amounts of data, much of which may never be examined in detail because analysing it requires time, computing resources, and human expertise. By performing initial analysis onboard, future satellites could reduce the volume of data that needs to be transmitted and processed on the ground.

Paul Lasserre, Loft Orbital’s head of AI, described the wider opportunity by saying: “If you have a VLM, you can have logic, like ‘monitor this border for me, and let me know when something is suspicious,’ and interact back and forth with the satellites.”

That represents a change from satellites acting primarily as remote cameras towards becoming active participants in monitoring and decision-making processes.

The technology could also help reduce delays. For example, rather than waiting for imagery to be downloaded and reviewed, operators could potentially receive alerts about significant events as they occur.

Potential Applications

The possible uses extend across both commercial and public-sector activities.

Loft Orbital already highlights applications including vessel detection, asset monitoring, border security, environmental tracking, wildfire detection, vegetation monitoring, and deforestation analysis. The company’s wider vision involves deploying AI applications directly in orbit rather than relying entirely on ground-based processing.

The company states that its AI-enabled infrastructure allows decision-makers to have “their questions answered in near real-time”.

Future systems could potentially monitor shipping routes, identify unusual activity around critical infrastructure, detect environmental changes, or support emergency response efforts following natural disasters.

Loft is also developing Altair, a planned ten-satellite AI-enabled constellation designed for near real-time monitoring and deployment of space-based AI applications.

Part Of A Bigger Change

The demonstration also points towards a broader trend within the space industry. For decades, satellites have primarily been designed to collect information and transmit it elsewhere for analysis. Increasingly powerful onboard processors are now making it possible for spacecraft to perform far more sophisticated tasks independently.

According to Loft Orbital, “Traditional satellites cannot keep up” with the pace of AI development, which is why the company is investing heavily in on-orbit computing and AI infrastructure.

Researchers involved in the project also see potential applications beyond Earth observation. NASA JPL has previously discussed how similar AI assistants could eventually help astronauts working on the Moon or Mars by providing interactive support without requiring constant communication with Earth.

What Does This Mean For Your Business?

The project really demonstrates how AI is increasingly moving closer to where data is generated rather than relying entirely on centralised data centres and cloud platforms. Similar trends are already emerging in manufacturing, transport, cyber security, healthcare, and industrial monitoring, where AI systems are being deployed directly at the edge rather than waiting for data to be sent elsewhere.

The satellite also highlights a change in how organisations may interact with technology. For example, rather than collecting information and analysing it later, future systems are increasingly being designed to understand objectives, identify relevant information, and proactively highlight what matters.

The result is not simply faster analysis, but it also represents a move towards autonomous systems that can act as intelligent assistants, helping people make decisions from vast amounts of data that would otherwise be impossible to process efficiently. As AI capabilities continue to improve, that model is likely to become increasingly common both in space and here on Earth.

Tesco is migrating approximately 40,000 servers away from VMware while simultaneously pursuing legal action against Broadcom for more than £100 million, in a dispute that highlights growing concerns about software licensing, vendor lock-in, and the risks of relying on critical technology platforms.

What Happened?

The dispute stems from Broadcom’s acquisition of VMware, one of the world’s largest providers of server virtualisation software, by US technology giant Broadcom in late 2023.

Before the takeover, Tesco had purchased perpetual VMware licences in 2021 that included software updates and support until 2026, together with an option to extend support arrangements until 2030. According to Tesco, those agreements formed part of its long-term technology planning for critical systems across the business.

Following the acquisition, however, Broadcom changed VMware’s licensing model and moved customers towards subscription-based offerings.

Tesco alleges that it was prevented from continuing with the support arrangements it expected under its existing agreements and instead faced significantly more expensive bundled subscription packages. The retailer is now pursuing legal action in the UK High Court against Broadcom, VMware entities, and other parties connected to the dispute.

Why Tesco Is Migrating

Perhaps the most striking aspect of the story is that Tesco is not waiting for the court case to conclude.

For example, the retailer has already begun a major programme to replace VMware across its estate and intends to complete the migration by the end of 2027. According to court filings, Tesco began migration efforts in 2025 after concluding that it could no longer depend on the outcome of the legal dispute.

It is worth noting here the sheer scale of this project. For example, migrating 40,000 servers is a complex undertaking that involves replacing core virtualisation infrastructure, retraining staff, testing alternative platforms, maintaining business continuity, and ensuring compatibility with a wide range of existing applications and services.

Tesco’s legal filings seem to suggest the company believes the operational disruption and migration costs are preferable to remaining dependent on its current VMware arrangements.

The Pricing Dispute

At the centre of the disagreement is the question of cost. According to Tesco’s legal filings, Broadcom offered a one-year VMware Cloud Foundation subscription that was substantially more expensive than the pricing Tesco expected under its previous renewal arrangements. The retailer claims some proposals represented increases of approximately 175 per cent compared with earlier pricing structures.

The dispute also extends beyond VMware. Tesco has claimed that software and support arrangements connected to CA Technologies mainframe products, which are also owned by Broadcom, became significantly more expensive following the acquisition.

In court filings, Tesco alleges that it has been forced to incur substantial costs procuring alternative products, hiring external specialists, and diverting internal resources towards accelerated migration projects.

A Wider VMware Backlash

The Tesco dispute has attracted quite a bit of attention because of its size, but it seems the issues at the heart of the case are affecting many VMware customers.

For example, since Broadcom completed the VMware acquisition, many organisations have reported significant changes to licensing, packaging, and purchasing arrangements. The company’s strategy has focused on simplifying VMware’s product portfolio and encouraging customers to adopt broader subscription bundles centred on VMware Cloud Foundation.

Supporters argue that this approach provides a more integrated platform and creates clearer product offerings. Critics, however, have expressed concerns about rising costs and reduced flexibility, particularly for organisations that previously relied on perpetual licences.

The situation has created opportunities for rival vendors. Companies including Nutanix, Microsoft, HPE, and other virtualisation providers have increasingly positioned themselves as alternatives for organisations seeking to reduce dependence on VMware.

The Challenge Of Depending On One Technology Platform

The Tesco dispute highlights a broader challenge facing organisations that depend heavily on a single technology platform.

Many businesses view software licences as long-term assets and build operational plans around assumptions about future support, pricing, and compatibility. Acquisitions can disrupt those assumptions, particularly when new owners adopt different commercial strategies.

The case also demonstrates how difficult it can be to move away from deeply embedded technology. Tesco’s migration involves tens of thousands of servers, third-party support arrangements, infrastructure redesign, application testing, and significant operational risk. Even for one of Britain’s largest retailers, replacing a critical technology platform is neither quick nor inexpensive.

Also, the fact that Tesco has decided to proceed with the migration while litigation continues appears to show the extent to which the relationship has broken down.

What Does This Mean For Your Business?

For businesses, the most important lesson is not whether Broadcom or Tesco ultimately prevails in court. The story really highlights the importance of understanding technology dependencies and having contingency plans for critical platforms. Software vendors can change ownership, licensing models, support arrangements, and commercial priorities far more quickly than organisations can replace core infrastructure.

It also demonstrates the value of regularly reviewing exit strategies. Many businesses carefully assess the benefits of adopting a platform but spend less time considering what would be involved if they needed to leave it.

Most organisations will never face a migration on the scale of Tesco’s 40,000-server project. However, the underlying challenge is familiar to businesses of all sizes. The more critical a technology becomes to daily operations, the more important it is to understand the risks associated with relying on a single supplier.

The Tesco case may ultimately become one of the most closely watched technology disputes of recent years because it raises a question that, when you buy software, how much control do you really retain if the company behind it changes?

New research shows that 43 per cent of MSPs and their customers experienced a cyber incident linked to a supplier or third-party vendor during the last year.

CyberSmart’s 2026 MSP Survey found that MSPs are increasingly being targeted because their access to customer systems can provide a route into multiple organisations. More than half of supply chain incidents involved the MSP as well as the customer.

The survey also found that only 45 per cent of MSPs continuously monitor third-party risk. CyberSmart CEO Jamie Akhtar warned that “a single weak link can have far-reaching consequences for customers, suppliers and partners”.

The findings come as MSPs prepare for the UK’s Cyber Security and Resilience Bill, which will increase scrutiny of supply chain security.

Businesses can reduce their exposure by reviewing supplier security, limiting third-party access, and monitoring supply chain risks on an ongoing basis.

The Trump administration has taken the unusual step of intervening in an environmental lawsuit against Elon Musk’s AI company xAI, arguing that the data centre at the centre of the dispute is so important to national security that it should be protected from legal action seeking to restrict its power supply.

Dilemma

The case highlights a growing sustainability dilemma facing the AI industry. For example, while artificial intelligence is increasingly being positioned as a tool for solving global challenges, its rapidly growing appetite for electricity is creating new environmental pressures, particularly as operators race to build ever-larger data centres.

Why The Government Has Intervened

The dispute in the U.S. centres on xAI’s Colossus AI facility in Mississippi, which relies on dozens of methane gas turbines to help power the infrastructure used to train and operate Grok, the company’s AI model.

The National Association for the Advancement of Colored People (NAACP), one of the oldest and largest civil rights organisations in the United States, filed the lawsuit. In this case, the NAACP’s Mississippi State Conference filed it. The lawsuit alleges that the turbines are operating without the permits required under the Clean Air Act and are contributing to air pollution that could affect nearby communities.

However, in a court filing submitted on behalf of the United States government, the Department of Justice argued that the lawsuit threatens “American national, economic, and energy security by seeking to shut off the power supply for artificial-intelligence innovation that supports the Department of War’s military operations.”

The filing seeks dismissal of the case and represents an unusually direct intervention by the federal government in support of a private technology company.

Why Grok Is Being Treated As A Strategic Asset

A key part of the government’s argument is that Grok has become integrated into sensitive national security operations. For example, according to a declaration submitted by Cameron Stanley, Chief Digital and Artificial Intelligence Officer at the Department of War (previously known as the Department of Defense), xAI’s Grok is “one of only four proprietary state-of-the-art (‘frontier’) AI models currently capable of supporting national security applications”.

The declaration also states that the Department relies on a specialised version known as Grok Gov Model and that it provides capabilities “found in no other frontier AI model”.

The filing claims that if the Mississippi facility were unable to continue operating at its current scale, the development and improvement of future Grok models could be affected, potentially impacting military and intelligence capabilities.

Whether or not the court ultimately accepts those arguments, the case demonstrates how rapidly advanced AI systems are being reclassified from commercial technology platforms into infrastructure that governments increasingly view as strategically important.

The Environmental Cost Of AI Growth

The environmental concerns at the heart of the case are difficult to ignore. For example, the turbines reportedly emit pollutants including nitrogen oxides and particulate matter, both of which have been linked to respiratory and cardiovascular health problems. Environmental groups argue that communities living near the facility should not bear the environmental cost of powering AI systems.

Also, the growth of AI is creating unprecedented demand for electricity. Modern AI models require vast numbers of processors working simultaneously, and those processors need enormous amounts of power.

The result is that many technology companies are now competing for access to electricity on a scale more commonly associated with heavy industry.

This creates an uncomfortable contradiction. Many of the same companies investing heavily in sustainability initiatives and clean technologies are simultaneously searching for whatever energy sources can support their rapidly expanding AI ambitions.

The Search For Cleaner Alternatives

The controversy also highlights why technology companies and investors are increasingly searching for lower-carbon energy sources capable of supporting AI’s growing power demands. One recent example is Critical Energy, a startup founded by a former SpaceX engineer that has raised $22 million to develop modular geothermal turbines for geothermal power plants. The company argues that geothermal energy could provide reliable, round-the-clock electricity for energy-hungry AI infrastructure years before many advanced nuclear projects become commercially available.

Projects such as these are attracting growing attention because geothermal energy can provide continuous power without the intermittency associated with solar or wind generation.

For the AI industry, that matters because data centres require power around the clock, making reliability almost as important as sustainability.

The challenge, however, is that many cleaner energy projects take years to deploy, while AI demand is growing today.

What Does This Mean For Your Organisation?

This case essentially offers an early glimpse of a debate that is likely to become increasingly common over the next decade. Governments want to lead in AI. Businesses want access to more powerful AI tools. At the same time, communities, regulators, and environmental groups are demanding that growth happens responsibly, and the lawsuit against xAI sits directly at the intersection of those competing priorities.

For organisations investing in AI, the wider lesson is that sustainability is becoming an infrastructure issue as much as a software issue. Questions about where AI runs, how it is powered, and what environmental impacts it creates are likely to become increasingly important alongside discussions about capability, productivity, and security.

The case also suggests that some governments are beginning to treat advanced AI infrastructure in much the same way as power stations, telecommunications networks, and defence assets. If that trend continues, future debates about AI may focus as much on energy policy and environmental impact as they do on the technology itself.

The political backdrop is also difficult to overlook in this particular case. Given Elon Musk’s close relationship with the Trump administration, some critics are likely to question whether the government’s intervention reflects purely national security concerns or whether political considerations may also have played a role. The administration maintains that its position is based on the strategic importance of the AI infrastructure involved.

The outcome of this particular lawsuit remains uncertain at this point. What is already clear, however, is that the race to build more powerful AI systems is creating difficult choices between economic growth, national security, environmental protection, and sustainable energy development, choices that governments, businesses, and communities will increasingly be forced to confront.

Microsoft’s new ‘Cowork’ feature in Copilot lets you assign tasks by simply describing the outcome, with Copilot creating a plan, using your Microsoft 365 data, and carrying out tasks across apps in the background while keeping you in control at every step.

[Note – To Watch This Video without glitches/interruptions, It may be best to download it first]

If Gemini AI prompts in Google Docs are getting in the way of your writing, there’s a quick way to turn them off and regain a cleaner, more focused workspace.

Why This Matters

Google has been steadily integrating Gemini into Google Docs, Sheets, Gmail, and other Workspace applications. While many users find the AI features useful, others prefer a cleaner writing environment without AI prompts appearing on screen.

Reducing these prompts can make documents feel less cluttered, help you focus on writing, and give you more control over when and how you use AI tools.

How To Hide The Gemini Bottom Bar In Google Docs

If you see a Gemini prompt box at the bottom of your document, you can hide it:

• Open your Google Doc.
• Click Gemini in the menu bar.
• Select Bottom Bar Preferences.
• Turn off the Gemini bottom bar.

This removes the AI prompt panel from the bottom of the document while allowing you to access Gemini manually if you need it later.

How To Reduce AI Features Across Google Workspace

You can also disable some Workspace smart features that power AI-driven suggestions and personalisation.

• Open Gmail.
• Click the Settings cog.
• Select See All Settings.
• Scroll down to Google Workspace Smart Features.
• Click Manage Workspace Smart Feature Settings.
• Turn off Smart Features in Google Workspace if you no longer want these features enabled.

This setting affects several Google Workspace applications and may reduce some AI-powered suggestions and prompts.

Things To Keep In Mind

Turning off Workspace smart features does not necessarily remove every Gemini capability from your Google account. Some Gemini features may still be available depending on your Google Workspace subscription, administrator settings, and the specific applications you use.

However, these settings can significantly reduce AI prompts and create a cleaner, less distracting workspace for users who prefer to write and work without constant AI assistance.

A Cleaner Way To Work

AI tools can be useful when you need them, but many people don’t want them appearing every time they open a document. Taking a few moments to adjust these settings can help create a simpler, more focused Google Docs experience while still allowing you to use Gemini when you choose to.

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.