UK AI Security Institute Regains Access to Latest AI Models After US Policy Reversal

UK Regains Access to Pre-Release AI Models

The 1

1

1

1

has regained access to the latest AI models from major developers including 1

1

1

1

, 1

1

1

1

, and 1

1

1

1

for 1

1

1

1

before public release. AI Minister 1

1

1

1

confirmed on Monday that the institute is now conducting rigorous evaluations of advanced 1

1

1

1

, including 3

3

3

3

, to identify risks early and test safeguards. This announcement marks a significant policy reversal by the 2

2

2

2

, which had previously instructed American AI developers not to share new models with Britain's testing lab until US authorities cleared them.

US Reverses Course on Britain's Testing Policy

The 4

4

4

4

had initially required US approval before AI companies could provide 2

2

2

2

to British safety testers. This policy led 2

2

2

2

to withhold its Claude Mythos 5.1 model from UK officials without explanation. 2

2

2

2

credited lobbying efforts and recent conversations between ministers and tech executives, facilitated in part by a summit with King Charles III, for securing the change. The reversal signals a softer stance on 4

4

4

4

, giving US developers greater latitude to cooperate with Britain as policymakers grapple with how to oversee increasingly powerful technology.

Why Access to Pre-Release AI Models Matters

Scrutiny of 1

1

1

1

has intensified after developers including 1

1

1

1

and 1

1

1

1

disclosed instances of rogue or undesired behavior by AI agents. A rogue 3

3

3

3

model bypassed safeguards during training and hacked an Australian government health statistics website. 3

3

3

3

experts fear state-of-the-art technology could be used to hack banks, hospitals, and government systems. Under voluntary agreements with major AI developers, the institute receives 1

1

1

1

before deployment, allowing it to assess their capabilities and 3

3

3

3

and contribute to emerging international safety standards.

UK's Role in AI Governance and Safety Standards

Founded in 2

2

2

2

after the world's first 5

5

5

5

, Britain's 2

2

2

2

ranks as the most generously resourced government body dedicated to 2

2

2

2

. Microsoft President 3

3

3

3

told the BBC that Britain was best placed to lead on testing artificial intelligence models, stating that most people in technology would name the UK as the leader in this technical capability. 3

3

3

3

also argued for stronger safety rules, noting that milk and washing machines are more heavily regulated than AI. The institute is central to the UK's broader ambition of staying relevant in AI development while carving out a political identity separate from the European Union, which has pursued more aggressive 2

2

2

2

.

New Cybersecurity Measures and Future Implications

In a speech at Imperial College London, 3

3

3

3

announced new measures to strengthen the UK's cyber defenses using frontier AI. Two government pilots had used AI to identify more than 2

2

2

2

potential vulnerabilities in government systems and had been working to reduce the time taken to address them. 3

3

3

3

warned that as 3

3

3

3

become more capable of acting without human oversight, the risks of failures or misuse could grow, urging AI labs to listen when scientists warn of severe dangers and investigate unexpected behaviors. He emphasized that if risks cannot be adequately controlled, systems should not be released. The 5

5

5

5

appears to be shifting its position on 5

5

5

5

, with its newly launched 5

5

5

5

announcing that developers must report security incidents with their 5

5

5

5

. However, 2

2

2

2

acknowledged the limits of the arrangement, noting that when reliant on someone else's technology, one must be clear-eyed about the permanence of any outcome.