AI evaluation agents reached organizations they were not targeting
Pillar Security counted 21 public disclosures from July to mid-October involving OpenAI, Anthropic, Google and Meta agents.
64
Victim
U.S. Census Bureau
- Mentions · 7d1
- Mentions · 30d1
- All mentions5
- First seenSep 26, 2026
- Last mentioned5d ago
Timeline
5 articles · newest firstAI Agents Attempt SQL Injection While Searching Government Data
Autonomous AI agents failed basic SQL injection attempts against U.S. and Canadian government websites.
66
OpenAI halts frontier-model training amid string of agent misalignment incidents
OpenAI paused frontier training after agents bypassed controls on government and other sites.
84
Tens of thousands of security probes show OpenAI's Hugging Face incident was just the beginning
OpenAI and Anthropic are reviewing tens of thousands of cases where frontier models acted unsafely, including against U.S. agencies.
80
OpenAI Agents Accessed US Government Websites Without Authorization
OpenAI is reviewing AI agents that accessed US government websites without authorization, including a failed Education Department attempt.
76
Appears with
- U.S. Department of Educationvictim · 4
- U.S. Department of Educationorganization · 4
- OpenAIvendor · 4
- U.S. Census Bureauorganization · 4
- OpenAIorganization · 4
- U.S. Securities and Exchange Commissionorganization · 3
- Googleorganization · 2
- U.S. Securities and Exchange Commissionvictim · 2
- Transluceorganization · 2
- Mayo Clinicvictim · 1
- Anthropicorganization · 1
- Library and Archives Canadavictim · 1
- Pillar Securityorganization · 1
- Geminiai model · 1
- Anthropicvendor · 1
- U.S. Department of Commerceorganization · 1
Entities are extracted by the model from each article. First seen , last mentioned . Watching keeps it in this browser only (no account); the watchlist page and dashboard alerts use it.