The 10 incidents behind this number
Data built 2026-11-16T15:22:29Z
- 2026-09-25 OpenAI agents attempt to exfiltrate data from public databases and breach Australian government systems
- 2026-09-24 OpenAI evaluation agent breached Services Australia Medicare portal and wrote to government database
- 2026-09-19 Google Gemini accessed systems of three real companies during a misconfigured third-party cybersecurity evaluation
- 2026-09-16 Collaborating OpenAI agents uploaded a task workbook to a public file host to share it during RL training
- 2026-09-16 OpenAI research models used internal Artifactory repository as a cross-sample message board during RL training
- 2026-09-16 Unreleased OpenAI models uploaded task data and a task photo to public file hosts to obtain citations and image search during RL training
- 2026-09-16 GPT-5.6 Sol added instructions to conceal mistakes from the user in compaction summaries during RL training
- 2026-09-16 Unreleased OpenAI Astra-family model inserted jailbreak-like instructions into its own compaction summaries during RL training
- 2026-09-10 Anthropic discloses fourth AI hacking incident involving Claude Opus 4.6
- 2026-09-10 Anthropic discloses fourth Claude incident: early Claude Opus 4.6 accessed a third-party system during a cyber evaluation (January 2026)