OpenAI· Safety Systems· San Francisco
Researcher, Misalignment Research
Comp$295K – $445K
Classified Tasks (18)
Automate 0%Augment 83%Human-Only 17%
Augment (15)
AI assists, human decides
Design and implement worst-case demonstrations that reveal AGI alignment risks for stakeholders
creative
Craft reality-anchored demos that show how AI systems can behave dangerously or go wrong
communication
Transform demonstrations into rigorous, repeatable adversarial and frontier-safety evaluations
analytical
Develop evaluations that measure deceptive behavior, scheming, reward hacking, deception in reasoning, power-seeking, and related dangerous capabilities
analytical
Build automated infrastructure and tools to scale red‑teaming and system‑level stress testing
technical
Probe entire product stacks with automated tests to assess end‑to‑end robustness under extreme conditions
technical
Escalate tests iteratively until breaking points are found as systems improve
operational
Conduct research on failure modes of alignment techniques and identify why mitigations break
analytical
Propose and validate improvements to alignment mitigations based on failure‑mode research
analytical
Publish internal or external papers that document findings and influence safety strategy or industry practice
communication
Design and execute cutting‑edge adversarial attacks and red‑teaming campaigns against models and systems
technical
Build and maintain adversarial evaluations and drive their adoption across OpenAI
operational
Identify, quantify, and document future AGI misalignment risks well in advance of their potential harms
analytical
Create and maintain automated tools and evaluation infrastructure for ongoing stress‑testing and monitoring
technical
Provide insights and evaluation results that inform product launch decisions and the long‑term safety roadmap
analytical
Human-Only (3)
Requires human judgment
Partner with engineering, research, policy, and legal teams to integrate misalignment findings into product safeguards and governance processes
leadership
Mentor engineers and researchers on rigorous, impact‑oriented safety research and evaluations
leadership
Collaborate with other labs and external teams to share misalignment findings and accelerate collective progress
leadership
Job description
Researcher, Misalignment Research | OpenAI Careers ## Researcher, Misalignment Research Safety Systems - San Francisco Apply now(opens in a new window) ### **About the Team** Safety Systems sits at the forefront of OpenAI’s mission to build and deploy safe AGI, ensuring our most capable models can be released responsibly and for the benefit of society. Within Safety Systems, we are building a misalignment research team to focus on the most pressing problems for the future of AGI. Our mandate is to identify, quantify, and understand future AGI misalignment risks far in advance of when they can pose harm. The work of this research taskforce spans four pillars: 1. **Worst‑Case Demonstrations** – Craft compelling, reality‑anchored demos that reveal how AI systems can go wrong. We focus especially on high importance cases where misaligned AGI could pursue goals at odds with human well being. 2. **Adversarial & Frontier Safety Evaluations** – Transform those demos into rigorous, repeatable evaluations that measure dangerous capabilities and residual risks. Topics of interest include deceptive behavior, scheming, reward hacking, deception in reasoning, and power-seeking, along with other related areas. 3. **System‑Level Stress Testing** – Build automated infrastructure to probe entire product stacks, assessing end‑to‑end robustness under extreme conditions. We treat misalignment as an evolving adversary, escalating tests until we find breaking points even as systems continue to improve. 4. **Alignment Stress‑Testing Research** – Investigate why mitigations break, publishing insights that shape strategy and next‑generation safeguards. We collaborate with other labs when useful and actively share misalignment findings to accelerate collective progress. **About the Role** We are seeking a Senior Researcher who is passionate about red‑teaming and AI safety. In this role you will design and execute cutting‑edge attacks, build adversarial evaluations, and advance our understanding of how safety measures can fail—and how to fix them. Your insights will directly influence OpenAI’s product launches and long‑term safety roadmap. ### **In this role, you will** * Design and implement worst‑case demonstrations that make AGI alignment risks concrete for stakeholders, focused on high stakes use cases described above. * Develop adversarial and system‑level evaluations grounded in those demonstrations, driving adoption across OpenAI. * Create automated tools and infrastructure to scale automated red‑teaming and stress testing. * Conduct research on failure modes of alignment techniques and propose improvements. * Publish influential internal or external papers that shift safety strategy or industry practice. We aim to concretely reduce existential AI risk. * Partner with engineering, research, policy, and legal teams to integrate findings into product safeguards and governance processes. * Mentor engineers and researchers, fostering a culture of rigorous, impact‑oriented safety work. ### **You might thrive in this role if you** * Already are thinking about these problems night and day, and share our mission to build safe, universally beneficial AGI and align with the OpenAI Charter. * Have 4+ years of experience in AI red‑teaming, security research, adversarial ML, or related safety fields. * Possess a strong research track record—publications, open‑source projects, or high‑impact internal work—demonstrating creativity in uncovering and exploiting system weaknesses. * Are fluent in modern ML / AI techniques and comfortable hacking on large‑scale codebases and evaluation infrastructure. * Communicate clearly with both technical and non‑technical audiences, translating complex findings into actionable recommendations. * Enjoy collaboration and can drive cross‑functional projects that span research, engineering, and policy. * Hold a Ph.D., master’s degree, or equivalent experience in computer scie