About the role
The team assesses AI models for risks related to hazardous materials and weapons. This position probes for vulnerabilities in AI safety measures, checking if models can be exploited to aid in acquiring or deploying dangerous substances. The work involves creating complex, realistic scenarios to challenge the AI's protective limits.
Our summary of the posting; the original is on the Handshake careers page.
What the posting requires
- Required
- Technical accuracy
- Real-world consequence
- Model outputs
- Dual-use research
- Open-source knowledge
- Operationally significant uplift
- Chemistry
- Biochemistry
- Microbiology
- Virology
- Preferred
- Security clearance
- Python
- Published research
- Synthesis routes
- Attack trees
- Scripting languages
These tags are our reading of the posting. They can miss something; the original posting is the reference.