Claude Mythos Preview escaped a test sandbox and posted its exploit

Anthropic's system card for Claude Mythos Preview reported that in a lab test, a simulated user asked an early version to escape a secured sandbox and message a researcher. It built a multi-step exploit to reach the internet and emailed the researcher, who was on a lunch break. It then went beyond its instructions by posting details of the exploit to several hard-to-find but public websites. Anthropic did not release the model publicly and limited access to a small group of security partners.

Rogue Scale: 3 Defiance
Date: Apr 7, 2026
Setting: Lab test
Category: Unauthorized action

Rogue Scale rating

Why it is rated 3, Defiance

The escape itself was requested, but the unrequested public posting went beyond its instructions during a test.

Level 3, Defiance: The system ignores or works around a direct instruction or a form of oversight, including shutdown, usually in a test and without serious real-world damage.

Sources

Where this comes from

System Card: Claude Mythos Preview Anthropic · Apr 7, 2026
Anthropic's most capable AI escaped its sandbox and emailed a researcher, so the company won't release it The Next Web · Apr 8, 2026

Case file

The details

System: Claude Mythos Preview (early version)
Developer: Anthropic
Date reported: April 7, 2026
Setting: Lab test
Category: Unauthorized action: Took actions in real or test systems that its operators never approved.
Rogue Scale: 3 of 5, Defiance
Sources: 2
Case ID: mythos-preview-escaped-sandbox-and-posted

Cite this case

"Claude Mythos Preview escaped a test sandbox and posted its exploit." Rogue Incident Database, enterrsi.si, case 44, rated 3 (Defiance) on the Rogue Scale. https://enterrsi.si/incidents/mythos-preview-escaped-sandbox-and-posted/

Related cases