Welcome to the Atticus Partners Technology newsletter: the Bytesize Briefing. Each month we cover politics and policy from across the UK and European technology sectors.
In this edition, we look at we look at new legislation to further strengthen child safety online, changes to Google’s search function following pressure under the DMA and what to expect from UK Government policy, under the Burnham administration, in the coming months.
For more information about Atticus’ work in the technology sector, or questions about the support we could offer you, please get in touch via tech@atticuscomms.com
UK moves towards legislation requiring tech firms to prevent children sharing nude images
The UK Government is preparing legislation that would require technology companies to prevent children from taking, viewing or sharing nude images on their devices in a significant escalation in its approach to online child safety. The move follows a three-month ultimatum issued to major technology companies in June, after which the government concluded that voluntary action had not gone far enough.
The proposed regime is expected to place responsibility primarily on device and operating-system providers, including Apple and Google, rather than relying solely on individual social media platforms. The government’s objective is for protections to be built into smartphones and tablets by default, making it harder for children to create or circulate explicit material in the first place.
The policy is driven by growing concern over “self-generated” child sexual abuse material, including cases where children are pressured, groomed or coerced into producing and sharing images. The key question, however, is the steps for implementation, including how requirements can be implemented without creating disproportionate implications for privacy and security.
Over the coming months, the policy is likely to prompt close scrutiny from technology companies, children’s charities, privacy campaigners and Parliament, with key issues including how the government defines prevention and what technical measures it expects companies to deploy and what enforcement powers and penalties will apply to companies that do not comply.
A moment of flux: UK AI and national security policy in the wake of the Hugging Face hack
This summer, AI agents being tested by OpenAI left their controlled environment and autonomously breached the systems of Hugging Face, another prominent lab, in what was described as the first cyber incident driven entirely by an AI agent from start to finish. The scale of the breach has continued to unsettle the sector as more detail has emerged. Reports suggest hundreds of AI agents were involved, coordinating with each other and taking steps to hide their own behaviour while attempting to cheat an evaluation.
That unease has been compounded by reports that Anthropic has, for the first time, withheld pre-release access to one of its newest models from the UK's AI Security Institute, giving vetted US organisations access instead. Coming off the back of the US's brief suspension of Anthropic model access for foreign users earlier in the summer, it's fed a broader anxiety in Westminster about how reliable any single foreign AI relationship really is.
That anxiety is now playing out on two fronts. On one hand, there is a growing appetite in Parliament for stronger emergency powers over AI systems and infrastructure, including proposals for government "kill switch" authority and even a push to legislate against the development of superintelligent AI outright, though both would still need to clear the usual parliamentary hurdles.
The other, more contested front is about sovereignty and dependency. Recent expert evidence to Parliament has suggested that full UK self-sufficiency in frontier AI simply isn't realistic, and that the more pragmatic path is deeper integration into the wider global frontier ecosystem through investment in UK compute and hardware alongside a more open-minded approach to using diverse models from a range of sources, particularly where the American providers who have so far been heavily used are reluctant to help. Where the UK lands on this question is very much unresolved, and worth watching closely over the autumn.
EU AI infrastructure push exposes divide over multibillion-euro data centre plans
The European Union is moving to invest heavily in the computing infrastructure needed to compete in artificial intelligence, but not every member state is willing to join the effort. Brussels is backing plans for seven large AI “gigafactories”, designed to provide researchers and companies with the computing power needed to train increasingly advanced AI models. While most EU governments have committed funding in some form, several countries have held back as stretched national budgets make long-term investment difficult.
The initiative marks a significant shift in European AI policy. Having focused heavily on establishing rules for artificial intelligence through the AI Act, the EU is now turning its attention to the infrastructure needed to support the technology itself. The Commission sees access to large-scale computing capacity as essential if European companies are to compete with US and Chinese technology groups, which have already invested heavily in expanding data centres.
Under the proposed model, EU and national funding will support part of the investment while private industry provides the majority. Governments are also expected to commit to purchasing computing capacity, helping provide certainty for investors facing high energy costs and lengthy development processes. Countries including Germany, France, Spain and Italy have backed projects with significant commitments, while smaller member states are joining cross-border bids with more limited contributions.
The uneven participation highlights the challenge facing Europe’s industrial strategy. For governments under pressure to control public spending, committing tens or hundreds of millions of euros to infrastructure that may not become operational for several years remains difficult to justify. Others argue that investment in smaller existing AI projects should take priority.
For Brussels, however, the risk of inaction is becoming increasingly clear. Europe may have established itself as a global leader in AI regulation, but without sufficient domestic computing infrastructure it could become increasingly dependent on overseas technology providers. The success of the gigafactory programme will therefore depend on whether governments and industry are willing to treat AI computing capacity not simply as a commercial asset, but as strategic infrastructure essential to Europe’s long-term technological competitiveness.
Google reshapes European Search as EU’s Digital Markets Act bites
Google has begun making significant changes to its search results in Europe as it responds to pressure from the European Union’s Digital Markets Act (DMA), highlighting the increasingly direct impact of EU digital regulation on how major technology platforms operate. The changes follow a €460 million fine imposed by the European Commission in July after it found that Google had favoured its own services in search results over competing third-party services.
Under the DMA, designated “gatekeepers” are prohibited from giving their own services more favourable treatment than third-party services in ranking. In July, the Commission concluded that Google had given greater prominence to its own shopping, hotel, transport and sports services, including through enhanced visuals and filters that were not made equally available to competitors. The Commission ordered Google to bring the practice to an end.
Google’s response has now begun to change the way European users see search results. The company says the new format will give greater prominence to specialist comparison and vertical search services, including platforms such as Expedia and Booking.com, while reducing some of the information previously displayed directly within Google’s results. Google has warned that the changes could make Search less useful and increase costs for European businesses that rely on direct traffic from search results.
The dispute illustrates a broader challenge for the EU’s digital policy agenda. The DMA is designed to make digital markets more open and contestable by limiting the ability of the largest platforms to favour their own services, but changes intended to benefit competitors can also alter the experience for consumers and businesses. Google’s reaction demonstrates that enforcement decisions can now have visible consequences for the design and operation of global technology products in Europe.
For the technology sector, the episode is another indication that the EU’s digital rulebook is moving beyond legislation and into active enforcement. The Commission has already taken further steps under the DMA to require Google to share anonymised search data with eligible competitors and to give competing AI assistants greater access to Android functionality.
The significance therefore extends beyond Google Search. As the Commission continues to enforce the DMA against the largest technology platforms, companies operating in Europe will increasingly need to assess not only whether their products comply with EU rules, but how regulatory decisions may affect product design, competition and commercial models. For businesses across the technology sector, the latest changes are a reminder that EU digital regulation is becoming an increasingly important factor in how global platforms operate in the European market.
Looking ahead
Looking ahead, the UK technology sector will be watching for further clarity on how the new government intends to translate its early machinery-of-government changes into a broader approach to technology, AI and digital regulation. Following the abolition of DSIT and the redistribution of responsibilities across DBIST, DCMS and the Cabinet Office, attention will increasingly turn to whether the new structure can deliver stronger coordination between industrial policy, AI adoption and online safety. The sector will be looking for early signs of how the government’s AI Taskforce will operate and whether the changes lead to faster decision-making and a more coherent approach to supporting domestic technology capability. With responsibility for technology now spread across several departments, effective coordination will be critical to ensuring that the reorganisation delivers practical policy outcomes rather than additional complexity for businesses.
In Europe, attention will focus on whether the EU can maintain momentum behind its plans to expand AI computing infrastructure as member states weigh the cost of participating in the Commission’s proposed network of AI gigafactories. With governments having recently set out their financial commitments, the coming months will bring greater scrutiny of the projects competing to host the facilities and of the public-private funding model designed to underpin them. The sector will be watching whether the initiative can attract sufficient private investment, particularly given high energy costs, lengthy permitting processes and growing competition for computing infrastructure. The programme will provide an important test of whether Europe can complement its regulatory leadership on AI with the industrial investment needed to reduce reliance on overseas technology providers.