UK Stand Firm on Testing OpenAI and Anthropic Models Before Launch
Britain has rejected a US request to halt the AI Safety Institute's access to the latest models prior to release, following the use of frontier models to detect over 1,000 potential vulnerabilities.
Britain has decided to maintain the AI Safety Institute's access to the most powerful models and test them before launch, responding to a previous US request that OpenAI and Anthropic refrain from releasing their new models to British testers, according to a report by The Next Web.
The White House had requested both companies last month to withhold their new models from testers in Britain.
AI Minister Kanishka Narayan said that two trials used advanced AI to discover more than 1,000 potential vulnerabilities in government systems.
A task force was formed to protect British government systems from AI-powered attacks. The decision reflects a race among nations to ensure the most advanced models are vetted prior to release, amid disagreements over who should be granted access to them.
What do these terms mean?
Frontier model: The most advanced AI models at a given time, whose sophisticated capabilities raise security concerns that require special inspection.
Pre-release testing: Making the model available to trusted entities for evaluation and risk discovery before public deployment.
Vulnerabilities: Flaws or weaknesses in systems that attackers could exploit to breach or disable them.
Weekly Newsletter
Read between the lines before everyone else. Decode the most important economic, tech, and decision-maker movements in the region.. in 5 minutes every Saturday.
