Daily AI Brief

During a UK state visit, US President Donald Trump and UK PM Keir Starmer signed a joint agreement to cooperate on AI development, quantum computing, and civil nuclear projects. Nvidia will deploy 120,000 GPUs across the UK — its largest European rollout to date.

US DOJ's Assistant Attorney General Gail Slater said the administration will focus on preventing anti-competitive control of data, key to AI model training. Recent court orders forcing Google to share data were cited as precedent.

In Washington DC, a Senate Judiciary subcommittee heard emotional testimony from parents, including over alleged suicides linked to interactions with AI chatbots. Chatbots from companies like Meta, Character.AI and others are under scrutiny for failing to restrict or prevent inappropriate content or interactions involving minors. Lawmakers are considering measures like mandatory safety protocols, age gating or even bans.

Bloomberg reports Oracle is negotiating to provide Meta with long-term AI compute capacity worth around $20 billion. The deal would make Oracle a major player in hyperscale AI cloud alongside AWS and Microsoft.

Reuters' Sustainable Switch column highlights the massive energy and water requirements of model training and inference, with calls for more transparency and green compute standards.

AI could raise trade value by nearly 40% by 2040, but without policy intervention, gains will disproportionately benefit advanced economies, warns the WTO's latest report.

Japan is developing a balanced approach to AI regulation that could serve as a model for New Zealand's own regulatory framework.

NZ academics and policy experts continue to advocate for clearer AI regulation covering privacy, health, and digital safety.