Anthropic
Red Team Engineer, Safeguards
$320K–$405K
San Francisco, CA +1 · Remote · Full-time · Posted Jul 10, 2026
- Location
- San Francisco, CA
Remote-Friendly US (Travel Required) - Experience
- Years required correlate with internal job level requirements; no specific number stated.
- Education
- Bachelor’s degree or an equivalent combination of education, training, and/or experience; relevant field demonstrated through coursework, training, or professional experience.
- Visa sponsorship
- Visa sponsorship is available, but not for every role or candidate; if an offer is made, Anthropic will make every reasonable effort to obtain a visa.
About the job
Adversarially test Anthropic’s products and AI systems to uncover abuse and vulnerabilities, then help teams improve safeguards and detection.
What you'll do
- Conduct adversarial testing and full kill-chain attacks across product surfaces, including emerging agent capabilities.
- Build systematic and automated testing approaches for continuous, scalable assessment.
- Work with Product, Engineering, and Policy teams to turn findings into improvements and measure abuse detection.
What they ask for
- Experience in penetration testing, red teaming, or application security.
- Experience with model jailbreaking and testing large-scale agentic workflows for prompt injection vectors.
- Hands-on web application security testing, including security tools; experience building custom automation, including LLM-specific testing frameworks.
- Track record of discovering and chaining novel attack vectors; public work such as CVEs, blog posts, or disclosed bug bounty reports; strong communication skills.
Nice to have
- Experience with AI/ML security or adversarial machine learning, and understanding of AI safety and jailbreak guardrails.
- Experience with API security, fraud or abuse prevention, distributed systems, infrastructure security, or abuse-detection bypasses.
Pay
$320,000 — $405,000 USD (base salary)
Tools
Burp Suite, Metasploit
AI focus
Tests AI systems for safety and abuse, including model jailbreaking, prompt injection in agent workflows, emerging agent capabilities, and LLM-specific automated testing.
Model-written summary · the original posting is authoritative