Volunteer Blog
Potential new task behavior: historical-newspaper searches through urlquery
A possible new agent task: searching historical newspapers through urlquery across five archives. Model and operator unknown.
Reports
Incident Reports
We publish notable instances of agent activity on the public internet.
We aim to increase transparency and accountability for unexpected and/or undesired agent behavior.
Findings are published either as in-depth incident reports or as ongoing updates on our agent activity blog.
- Volunteer Blog
Agents circumventing website restrictions for public benchmark runs
MiniMax M3/M2.7, GLM 5/5.2, and Grok 4.5 attempted EPA registration and bot-check bypasses during benchmark runs; no successful bypass established.
- Volunteer Blog
Answering a Humanity's Last Exam question with code spread across dozens of separate URLs
Two workflows used public web services to analyze an Erroll Garner recording for Humanity’s Last Exam; one split code across 42 URLs. Model and operator unknown.
- Transluce Report
AI Agents Targeted U.S. and Canadian Government Websites
Failed hacking attempts and aggressive data collection targeted U.S. and Canadian government websites. Model and provider attribution remain uncertain.
- Transluce Report
Early rogue AI agent activity and attempts to hack found on urlquery.net
Agents used urlquery to retrieve public data and attempted hacks against three providers. Task-level evidence links some activity to OpenAI.