The Mythos model is reported to have been accessed by unauthorized users. Anthropic has previously stated that its technology is extremely powerful.

According to a source familiar with the matter and documents reviewed by the media, a few unauthorized users have obtained access to Anthropic PBC's new generation Mythos AI model. Anthropic has stated that the technology of this model is extremely powerful and could even be used to launch dangerous cyber attacks. A person who requested anonymity out of fear of retaliation revealed that on the same day Anthropic announced plans to open testing of the Mythos model to a few companies, a small group of users on a private online forum successfully accessed Mythos. The source said that this group has been regularly using Mythos since then, but not for cybersecurity purposes. The source also provided screenshots and live model demonstrations to corroborate this information. This unauthorized access, which was previously undisclosed, highlights the challenges Anthropic faces in completely preventing the spread of this potentially most powerful and dangerous technology beyond authorized partners. This incident has raised more questions: are there others using Mythos without permission, and what are their intentions.