Frontier AI Safety Threat Modeller

Google LLC; New York (NY); On-site; USD 174,000 - 252,000; Full time

Benefits offered by this job

Equity

Benefits

Job summary

Google DeepMind’s CBRNE team seeks a Threat Modeller SME to build and advance threat models for dual-use risks in advanced AI systems. You will work closely with the Threat Modeling Lead to calibrate risk and support safe model deployments.

The role requires absorbing complex scientific data, identifying critical blind spots, and delivering rapid, actionable analyses in a fast-paced environment. You will contribute to frontier safety with real-world threat intelligence.

Qualifications

• PhD in Science, Engineering, Data Science or equivalent practical experience.
• 3 years of experience designing qualitative and quantitative models.
• 1 year of experience working within a national laboratory, research organization, government defense organization, military intelligence unit, or specialized research organization.

Responsibilities

• Refine and maintain the threat modelling framework and integrate into FSF review processes.
• Partner cross-functionally to design evaluations identifying CBRNe risks in high-capability AI models.
• Collaborate to determine if deployment safeguards are adequate for high-capability models.
• Engage with external stakeholders including governmental entities and industry bodies.
• Monitor competitor landscape and emerging dual-use methodologies.

Skills

Qualitative & quantitative models

Threat modeling

Education

PhD in Science/Engineering/Data Science

Job description

Google DeepMind’s CBRNE team seeks a Threat Modeller SME to build and advance threat models for dual-use risks in advanced AI systems. You will work closely with the Threat Modeling Lead to calibrate risk and support safe model deployments.

The role requires absorbing complex scientific data, identifying critical blind spots, and delivering rapid, actionable analyses in a fast-paced environment. You will contribute to frontier safety with real-world threat intelligence.