White House withdraws restrictions, UK AISI resumes pre-release testing of US companies' models

1. 13:40
   网页Techmeme· Discussion

   Sources: the US restores the UK AISI's access to the latest US models for safety testing, after ordering OpenAI and Anthropic to stop sharing models last month (Tony Diver/Telegraph)(opens the original post in a new tab)
2. 07:39
   网页IT之家· Discussion

   UK to strengthen cyber defenses, AI minister calls on AI companies to increase security safeguards(opens the original post in a new tab)
3. 07:01
   The Guardian · 人工智能

   The UK will strengthen cyber defense, and the AI minister calls on frontier AI companies to increase safety investment

   UK AI minister Kanishka Narayan will announce strengthened UK cyber defenses and call on companies developing frontier AI to increase safety investment, saying the UK cannot rely entirely on tech companies' self-regulation. Geoffrey Irving, former chief scientist at the UK AI Security Institute, told the Guardian that AI companies lack security measures to prevent hostile states from stealing models, citing the case of OpenAI being "breached" in a test program that rewarded ethical hackers.

1. Time not given
   Reuters

   The Daily Telegraph reports that the US has withdrawn restrictions and no longer opposes US AI developers giving the UK AISI early access to models

   According to The Daily Telegraph, the Trump administration withdrew its previous decision requiring OpenAI, Anthropic, and other developers not to share new models with the UK AISI before release, allowing US AI developers to open models to the UK government for safety testing in advance. The report says this change indicates a softening of the Trump administration's stance on international AI safety efforts, leaving more room for US developers to cooperate with the UK. Spokespeople for Trump, OpenAI, and Anthropic did not immediately respond to Reuters' requests for comment. The UK AISI was established after the first AI Safety Summit in 2023 and is responsible for testing the capabilities and misuse risks of the most advanced AI models. Trump's newly formed AI task force said on Friday that developers must report safety incidents involving their AI models.

2. Time not given
   The TelegraphSelected

   White House reverses course, UK AISI regains pre-release testing rights for models from three US labs

   The White House has reversed its decision requiring U.S. AI companies to stop sharing models with the UK AISI, and the UK AI Safety Institute will resume pre-release safety testing access to the latest models from Anthropic, Google DeepMind, and OpenAI. Last month, after the White House's Office of the National Cyber Director issued the requirement, Anthropic restricted UK officials from reviewing its new Claude Mythos 5.1 model; according to The Telegraph, the decision was reversed after lobbying by AI minister Kanishka Narayan. Narayan will also announce a series of measures in a Monday speech, including amending the Cybersecurity and Resilience Act to strengthen requirements for companies undertaking critical national infrastructure projects to report AI-related safety incidents, and increasing government use of frontier AI to protect Whitehall departments, as well as adding resources for agencies like NCSC to investigate and respond to incidents.

1. Date not given
   UK AISI

   UK strengthens national defenses against growing AI risks

   The UK will adopt a package of measures to strengthen national defenses to better address the risks posed by increasingly capable AI.

Current attention70·Peak in the comparable period70（Oct 12 19:00）·Change over the comparable past 24 hours –