Red Team Engineer, Safeguards

About the role

Anthropic’s Safeguards team is looking for a Red Team Engineer to proactively uncover vulnerabilities in our AI‑driven products. You will adopt an adversarial mindset to simulate sophisticated threat actors, ensuring that our systems remain safe and trustworthy before they are exposed to real‑world abuse.

Key responsibilities

• Design and execute comprehensive adversarial tests across all product surfaces, creating multi‑vector attack scenarios.
• Research and prototype novel testing methods for emerging AI capabilities such as agentic systems and tool use.
• Develop “full kill‑chain” simulations that mimic real‑world malicious objectives.
• Build and maintain systematic testing methodologies and automated frameworks for continuous, large‑scale assessment.
• Collaborate with Product, Engineering, and Policy teams to translate findings into concrete security improvements.
• Help define metrics to measure detection effectiveness of novel abuse patterns.

Required profile

• Proven experience in penetration testing, red teaming, or application security.
• Hands‑on work with model jailbreaking and large‑scale agentic workflow testing.
• Track record of discovering novel attack vectors and publishing CVEs, blog posts, or bug‑bounty reports.
• Strong written and verbal communication skills for cross‑functional collaboration.

Required skills

• Penetration testing and red teaming
• Application security
• Model jailbreaking
• Web application security
• Security testing tools (e.g., Burp Suite, Metasploit)
• Custom scripting and automation
• LLM‑specific testing frameworks