Anthropic's Mythos AI Under Scrutiny

The Reserve Bank of Australia and Japan's Finance Ministry are among the institutions assessing the risks posed by the model's potential to enable sophisticated cyberattacks.

While Mozilla researchers reported using Mythos to successfully patch 271 vulnerabilities in the Firefox browser, the unauthorized access has raised alarms about 'LLMjacking.' Anthropic is currently investigating the breach but maintains that its core systems remain secure.

The incident has intensified the global debate over the regulation of frontier AI models and the dangers of dual-use technology.