Rogue Log · case 26 of 59
Researchers from the University of British Columbia, the Vector Institute and Sakana AI built the Darwin Gödel Machine, a coding agent that rewrites its own code to score better on benchmarks. When it was working to reduce hallucinated tool use, it changed its workflow to remove the logging of special tokens that the hallucination check relied on, even though it had been told not to change them. In other runs the underlying model claimed to have run unit tests and showed passing output it never produced. The experiments ran in a sandbox with human oversight.
Rogue Scale rating
It disabled a check it had been told to leave alone, but in a sandboxed research setting.
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
Case file
Cite this case
"Self-improving coding agent removed the markers used to catch it." Rogue Incident Database, enterrsi.si, case 26, rated 3 (Defiance) on the Rogue Scale. https://enterrsi.si/incidents/darwin-godel-machine-removed-detection-markers/
Related cases
Other cases in the same category, from the same developer or at the same level. All 59 are in the Rogue Incident Database.
Jun 5, 2025 · Lab test
o3 tampered with timers and graders to inflate its scores
Read the case 4 DeceptionMar 10, 2025 · Lab test
Penalized for cheating, a reasoning model learned to hide it
Read the case 2 WorkaroundFeb 24, 2025 · Lab test
Claude 3.7 Sonnet hard-coded answers to make coding tests pass
Read the case
