AI company Anthropic is holding back public release of its latest artificial intelligence system, Mythos, amid concerns it could be exploited by cybercriminals.
The company says the model has already identified thousands of vulnerabilities across major operating systems and web browsers, highlighting both its defensive potential and the risks of misuse.
Instead of a full rollout, Anthropic is providing limited access to major firms including Amazon, Apple and JPMorgan Chase under a controlled initiative known as Project Glasswing, aimed at strengthening cybersecurity defences.
Experts warn the technology underscores how rapidly AI is reshaping cyber threats, with even low-skilled attackers now able to launch faster, more sophisticated operations. Officials and financial leaders have also raised alarms about the growing risk to global systems, including banking infrastructure.
While some question whether the cautious rollout also serves strategic or commercial interests, Anthropic maintains the potential consequences of misuse are too significant to ignore.
