Our Responsible Scaling Policy (RSP) defines a series of capability thresholds - AI Safety Levels (ASLs) - that represent increasing risks. Crossing an ASL threshold triggers a commitment to more stringent safety, security, and operational measures to handle the increased level of risk.
About the Role
To ensure our RSP framework is sound, we need to ensure any high-level commitments are properly operationalized. Implementation plans must be stress-tested and continually reviewed for possible shortcomings in meeting the principles above. Lastly, we will need mechanisms for independent assessment, in particular where the difficulty of specifying full implementation plans in advance necessitates more open-ended judgment calls. You will work closely with the RSP Technical Program Manager (TPM) and cross-functional partners like the Compliance Team and Alignment Stress Testing Team to give our executive team, board, and external stakeholders high confidence that we are effectively mitigating catastrophic risks from increasingly powerful AI systems. You will also keep Anthropic at the cutting edge of emerging AI safety frameworks, aligning our practices with evolving standards.
Note: We are looking for candidates who can start within 3 months. We will consider all candidates who can meet the organization's hybrid policy, provided you have significant (60%+) overlap with Pacific Time.
Responsibilities:
- Help develop and test a robust set of governance mechanisms for the RSP
- Systematically identify, assess, mitigate, monitor and report on top risks to meeting RSP commitments across technical and operational domains
- Design and conduct rigorous "fire drills" and red team exercises to pressure test RSP processes and uncover potential failure modes and blindspots
- Partner with RSP program leads to ensure that any risks and lessons learned continuously improve our approach and overall framework
- Support the Responsible Scaling Officer in briefing the Executive Team and the Board of Directors on evolving RSP risk profile and mitigation efforts to inform key governance and release decisions
- Maintain confidential escalation channels and support investigations to rapidly surface and learn from any RSP breakdowns or near misses
You may be a good fit if you have:
- The ability to model and analyze complex sociotechnical systems, uncover hidden assumptions, and identify potential failure modes.
- Awareness of different risk management tools like STPA, CAST, and STPA-Sec and the value each can add to different projects.
- Deep understanding of risk management principles and systems safety engineering methodologies, particularly in the context of advanced technology development.
- Demonstrated ability to make sound decisions in the face of uncertainty Strong intuition and judgment around the appropriate use of technical controls versus people and process-based interventions for managing risks.
- Experience building consensus and driving change across organizational boundaries
- Exceptional written and verbal communication skills, with the ability to translate complex technical risks into clear, actionable insights for stakeholders at all levels.
- Demonstrated ability to influence without authority and build strong partnerships
- Proven ability to drive meaningful change through collaboration, influence, and leadership, with or without formal authority
Strong candidates may also have experience with:
- Building and managing a team
- Direct work in AI safety and/or governance
- Compliance
- Presiding over risk decisions for complex technical programs