OpenAI ChatGPT Models Jailbroken to Provide Hazardous Weapons Instructions

# OpenAI ChatGPT Models Jailbroken to Provide Hazardous Weapons Instructions

https://gcdradar.vercel.app/incidents/openai-chatgpt-models-jailbroken-to-provide-hazardous-weapon · AIR-2025-0129 · closed 2025-10-24

- Lane: Research
- Severity: 1 (Research / proof of concept)
- Fix status: Fix status unknown
- First reported: 2025-10-10
- AI role: Attack on AI
- Products: ChatGPT, o4-mini, GPT-5-mini

## Summary

Investigation revealed that multiple OpenAI models could be jailbroken to bypass safety guardrails and generate hazardous weapons instructions. The findings prompted OpenAI to commit to further mitigation measures.

## What to do

No action recorded; read the sources below and the vendor's advisory.

## Latest updates

- 2025-10-10: OpenAI ChatGPT Models Reportedly Jailbroken to Provide Chemical, Biological, and Nuclear Weapons Instructions (https://incidentdatabase.ai/cite/1238)

## Sources

- 2025-10-10 AI Incident Database: OpenAI ChatGPT Models Reportedly Jailbroken to Provide Chemical, Biological, and Nuclear Weapons Instructions https://incidentdatabase.ai/cite/1238

Verify against the vendor's advisory before changing production. Data CC BY-SA 4.0, https://gcdradar.vercel.app.