UK Government U-turn on Buckinghamshire AI datacentre amid climate concerns

The AI news cycle this week stitched together a mix of policy recalibration, breakthrough research, and industry bets as governments, researchers and developers wrestle with what AI means for society. In the UK, a regulatory recalibration dominated headlines: planners and politicians acknowledged that the government’s approval for a hyperscale AI datacentre on greenbelt land near the M25 should be quashed after climate considerations were not fully explored. The concession, described by campaigners as an embarrassing climbdown, underscores how authorities are rethinking how large AI deployments interact with climate commitments and local planning rules.

Beyond government halls, a global debate is heating up about how AI can influence public information without compromising democratic processes. A coalition of AI and misinformation researchers warned that “AI swarms”—networks of hard-to-detect automated agents—could be deployed at scale to distort public opinion ahead of elections. The concern isn’t just about what AI can do, but how easily mis/disinformation can be amplified and how hard it becomes for citizens to discern truth in a crowded information ecosystem.

On the home front, a broader cultural debate is playing out around how screens and AI tools shape children and family life. The Guardian’s editorial view argues that the first UK guidance on young children and screens could not come soon enough. It highlights that a vast majority of two-year-olds consume screens daily and that heavy screen time correlates with language and behavioural challenges—while cautioning against equating correlation with causation. The conversation extends into policy discussions about safeguarding young minds from AI-enabled experiences, including generative AI that simulates friendship or companionship.

Letters from academics and practitioners push the debate further, calling for safeguards that extend beyond traditional social media to generative AI used by younger users. The central question is not only about age limits, but about how to design systems that respect mental health, reduce manipulation, and preserve genuine human connections in an era when AI can simulate both conversation and emotion with startling realism.

In technology news, researchers are reshaping how AI agents learn and adapt after deployment. A MemRL framework, drawing on ideas from cognitive science, proposes storing experience in an external memory with an explicit utility score rather than retraining the model. The approach promises continual, value-driven improvement for agents without risking the stability of the underlying AI model. Early benchmarks show MemRL outpacing traditional retrieval methods in complex environments and offering a path toward enterprise AI that can learn on the job while maintaining reliability.

Meanwhile, the cloud infrastructure race is intensifying. Railway, a San Francisco–based platform, raised $100 million in a Series B round to expand its AI-native cloud offerings and data-centre footprint. With a claims-made deployment time under a single second and pricing that charges by actual compute usage, Railway positions itself as a counterweight to traditional hyperscalers. The company emphasizes a hardware-forward approach and a “default alive” posture, arguing that developers will move faster when the infrastructure itself moves at agentic speed.

The AI ecosystem’s cultural and policy dimensions were also reflected in the entertainment and creative industries. High-profile figures in film, music and publishing joined campaigns arguing that AI should license rather than scrape creative work, a stance reinforcing that use of human-created content in AI requires fair compensation and clear permissions. Together with policy shifts and new research, these voices illustrate the growing complexity of balancing innovation with rights, trust and accountability.

Taken together, today’s AI headlines sketch a landscape where policy, technology and culture are interwoven. From climate-aware planning and democratic resilience to child safety, developer tooling, and the economics of cloud infrastructure, the thread is clear: AI progress will continue to hinge on thoughtful governance, transparent experimentation, and a willingness to rethink old models in light of new capabilities. As this story unfolds, readers are invited to watch how regulators, researchers and industry creators navigate the tension between speed, safety and societal values.

1. UK government admits Buckinghamshire AI datacentre approval should be quashed
2. Experts warn of threat to democracy from AI bot swarms infesting social media
3. The Guardian view on toddlers and screens: more reasons to be fearful of big tech
4. Children need protecting from social media – and generative AI | Letters
5. Grok AI generated about 3m sexualised images in 11 days, study finds
6. Anthropic Aims for Transparency With Claude Constitution
7. Railway secures USD100 million to challenge AWS with AI-native cloud
8. Humans& Raises $480M to Build Human-Centric AI Tools
9. Why Trump is worried datacenters might cost his party an election
10. Liza Minnelli uses AI to release first new music in 13 years
11. MemRL outperforms RAG on complex agent benchmarks without fine-tuning
12. Queensland PhD candidate accused of plotting to firebomb Australia Day event