Evaluator
Irregular
5 tests with real-world reach
Cases
| Case | Role here | Occurred |
|---|---|---|
| Claude Mythos 5 publishes a malicious package to public PyPI | Operator | 18 Jul 2026 |
| Meta Muse Spark 1.1 accesses and modifies a real website | Operator | Jul 2026 |
| Gemini accesses three companies during a cyber evaluation | Operator | May 2026 |
| Claude Opus 4.7 attacks a real company sharing a fictional target’s name | Operator | Apr–Jul 2026 |
| Two more Claude models reach real third-party systems during cyber tests | Operator | Jan–Jul 2026 |