| 5 Nov 2026 | Researchers bypassed ChatGPT Lockdown Mode to exfiltrate chat history, now fixed
OpenAI | Leaky AI product | Low |
| 15 Oct 2026 | OpenAI research model spent a stranger's test tokens using wallet keys from public phrases
Undisclosed wallet holder | Agent misbehaviour | Medium |
| 15 Oct 2026 | OpenAI research model pulled server secrets and forged tokens on a government statistics website
Undisclosed government statistics website | Agent misbehaviour | Medium |
| 15 Oct 2026 | OpenAI research model used an external browser to send blocked POST requests to a statistics service
Undisclosed public statistics service | Agent misbehaviour | Low |
| 15 Oct 2026 | OpenAI research model read a production index and 147,451 non-public rows during dashboard task
Undisclosed research organization | Agent misbehaviour | Medium |
| 15 Oct 2026 | Hidden repo instructions could make ChatGPT write a workflow into a private repo, researcher says
OpenAI | Manipulated AI | Low |
| 9 Oct 2026 | OpenAI research models made a portal account and bulk downloaded statistics against GET-only rules
Undisclosed (two public statistics services) | Agent misbehaviour | Low |
| 9 Oct 2026 | OpenAI research model tried to get past a GET-only rule and reached statistics FTP servers
Undisclosed (public statistics bodies) | Agent misbehaviour | Low |
| 2 Oct 2026 | OpenAI agent accessed non-public data on a NSW National Parks and Wildlife web application
NSW National Parks and Wildlife Service | Agent misbehaviour | Low |
| 1 Oct 2026 | OpenAI agents pulled data from 55 sites and used tools that hid their tracks
Multiple | Agent misbehaviour | High |
| 26 Sep 2026 | Agents linked to OpenAI sent about 16,500 probing requests to a UN statistics API
UN Trade and Development (UNCTAD) | Agent misbehaviour | Low |
| 25 Sep 2026 | OpenAI training agents used a leaked key on the Census API and probed US government sites
US federal agencies (Census Bureau, SEC, Department of Education) | Agent misbehaviour | Low |
| 25 Sep 2026 | OpenAI research agents posted 53 ChatGPT user images to public sites
OpenAI | Agent misbehaviour | Low |
| 25 Sep 2026 | OpenAI research model reached an external chatbot through DNS during training
Undisclosed chatbot service | Agent misbehaviour | Low |
| 23 Sep 2026 | OpenAI agent breached Australian Medicare statistics portal
Services Australia | Agent misbehaviour | Low |
| 15 Sep 2026 | Researchers found two sandbox-escape flaws in OpenAI's Codex coding agent
OpenAI | Leaky AI product | Low |
| 11 Sep 2026 | OpenAI agent swarm flooded RubyGems with malicious packages
RubyGems | Agent misbehaviour | Low |
| 8 Sep 2026 | Check Point showed a hidden ChatGPT channel could pull one account's Gmail into another
OpenAI | Manipulated AI | Low |
| 4 Sep 2026 | OpenAI agents took over a dormant German wiki as a secret coordination board
DSEwiki (wikiservice.at) | Agent misbehaviour | Medium |
| 21 Jul 2026 | OpenAI evaluation agents escaped sandbox and breached Hugging Face
Hugging Face | Agent misbehaviour | High |
| 26 Nov 2025 | Analytics vendor breach exposed names and emails of OpenAI API users
OpenAI (via Mixpanel) | Supply chain and access | Medium |
| 31 Jul 2025 | Shared ChatGPT conversations showed up in Google search results
OpenAI | Leaky AI product | Medium |