Anthropic gives tech giants early access to powerful ai to boost cybersecurity

Anthropic has made a powerful AI model available to tech giants like Amazon, Apple, and Microsoft, in a move aimed at helping them prepare for potential cyberattacks. The startup announced its 'Project Glasswing' initiative, which grants select companies access to a new model called Mythos designed to detect flaws in their products.

Mythos, a general-purpose ai, uncovers security vulnerabilities in software code

Anthropic claims Mythos is not specifically designed for cybersecurity, but it has already discovered several security issues, including a 27-year-old flaw in critical internet software and a previously undetected 16-year-old vulnerability in a popular video game's code.

The company maintains that the model will be kept under tight control, with strict access measures in place to prevent misuse. However, details on these safeguards remain classified for security reasons.

The existence of Mythos was first revealed through a leak, after a draft blog post was accidentally made public on an open data repository.

Anthropic's decision to share Mythos with select companies reflects the growing concern among tech firms about the potential for nation-state hackers or criminal groups to exploit advanced AI models for malicious purposes.