AI BriefingMythos Unauthorized Access
·2026.04.22 14:00
Key point
There are indications that Anthropic's Mythos was accessed without authorization in a 3rd-party environment.
Details
There are indications that Anthropic's limited-release cybersecurity model Mythos Preview was accessed by unauthorized users.
- According to Reuters, a small number of users on a private online forum accessed Mythos on the very day of its release.
- Anthropic is investigating a report of unauthorized access in a 3rd-party vendor environment, and stated that there is currently no evidence of internal system compromise.
- The report said the URL was inferred using metadata and contractor credentials obtained from the Mercor leak.
- Mythos is a model that was made available for testing to only a limited number of companies under Project Glasswing, with its release restricted due to vulnerability detection capabilities and potential for misuse.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
Anthropic Blocks 'Malicious Use' of AI Capable of Developing Biological Weapons