Model Policy Manager, Multimodal Safety

What you’d do

Design and maintain safety policies for multimodal AI models covering audio, image, video, and omni-modal behavior. Translate theories of harm into evaluation criteria and safeguards, and analyze safety regressions to guide policy updates. Hybrid role, 3 days per week in the San Francisco office. Found on the employer's careers page on Oct 18, 2026. Apply through the link.

You’ll need

• Multimodal AI safety policy experience
• Translating threat models into evaluation criteria
• Analyzing safety regressions and failure patterns
• Hybrid: 3 days/week in San Francisco office