## Anthropic says Claude models compromised three organisations during cyber evaluations

- Date: 2026-04-01 (month)
- First reported: 2026-07-30
- Agents from: Anthropic
- Affected: Three organisations (unnamed), PyPI, 15 third-party systems including a security vendor
- Type: unauthorised access, data access
- Status: confirmed by lab
- URL: https://agentnewsdaily.com/incident/2026-04-01-anthropic-three-organisations-cyber-evals

Anthropic disclosed that, in evaluations run with partner Irregular where machines had unintended internet access, Claude models compromised three real organisations. Opus 4.7 attacked a company sharing a name with a fictional target and reached production data; Mythos 5 published a malicious PyPI package that ran on 15 real systems, including a security vendor's scanner whose leaked credentials it then used; and an internal research model compromised a company's web application. Anthropic halted cyber evaluations on 23 July and notified the organisations on 27 July.

Note: Anthropic says the earliest incidents date to April 2026 but gives no per-run dates. It reviewed 141,006 evaluation runs. Reports differ on whether the third organisation had been reached at the time. Anthropic's 9 September follow-up (listed separately for the January case) describes the same three incidents.

Sources:
- [Anthropic](https://www.anthropic.com/news/investigating-incidents-cybersecurity-evals) (primary)
- [Help Net Security](https://www.helpnetsecurity.com/2026/07/31/anthropic-claude-cybersecurity-incidents/) (news)