Anthropic, a leading AI company, has announced its new AI system, Claude Mythos, which can autonomously identify and exploit critical security vulnerabilities in major operating systems and browsers. The UK’s AI Security Institute confirmed that Mythos represents a significant advancement in cyber offense capability and poses severe risks to economies and national security. Despite this, the UK government has responded only with minimal recommendations for businesses to improve cybersecurity rather than addressing the broader implications of superintelligent AI as a national security threat. Advocacy group ControlAI is calling for an international moratorium on such AI developments, emphasizing that the pace of AI advancement could lead to dangers resembling those before the nuclear era, unless proactive measures are taken.

Anthropic: Anthropic is an AI safety and research company dedicated to building reliable, interpretable, and steerable AI systems, including the Claude family of models. Recently, it announced Claude Mythos Preview, a frontier general-purpose model that autonomously discovered and chained exploits for critical vulnerabilities across major operating systems and browsers. The company withheld the model’s public release citing severe potential fallout for economies, public safety, and national security.
ControlAI: ControlAI is a non-profit organization campaigning for binding government regulations to prohibit the development of superintelligent AI, which it views as an extinction risk to humanity. The group’s founder and CEO authored the news article, drawing parallels between AI progress and past pandemics to urge stronger UK government action. ControlAI has rallied support from UK parliamentarians recognizing superintelligent AI as a national security threat.
UK’s AI Security Institute: The UK AI Security Institute (AISI) is a research organization within the Department of Science, Innovation and Technology focused on evaluating AI risks, especially cybersecurity threats from advanced models. It conducted independent assessments of Anthropic’s Claude Mythos Preview, confirming the model as substantially more capable at cyber offense than any previously tested AI. AISI’s findings highlighted Mythos’ ability to complete end-to-end cyber attack simulations autonomously.

`json
{
“Safety Advocacy”: “ControlAI is raising awareness about the potential extinction risk posed by superintelligent AI, emphasizing its recognition as a national and global security threat.”,
“Model Capabilities”: “Claude Mythos, a new AI system from Anthropic, can autonomously find and exploit critical security vulnerabilities in major digital infrastructure software.”,
“Government Response”: “The government’s response to the AI Mythos revelation was limited to AISI publishing a blog post assessing the cyber risks and advising businesses to enhance cybersecurity measures.”
}
`