Founding , DC Office; Science Communications, Engagement and Policy Research

Apollo Research Governance Team

Apollo Research is a mission-driven startup committed to reducing the threats from scheming frontier AI systems. We work with and are trusted by every frontier AI lab. We do fundamental research, test their models before deployment, and collaborate with them on scheming mitigations. The Governance team conducts research and provides advice on the institutional, legal and strategic implications of AI scheming and loss of control threats, working closely alongside our Science of Scheming, Evaluations, and Monitoring teams.

The Governance team works on a bet that the national security interventions the U.S. government enacts and implements over the next 15 months, through legislation, procurement and verifiable procedures for testing, monitoring, controlling and pacing frontier AI, will be among the most significant contributions to our collective ability to stay in control of increasingly advanced AI systems.

That means our DC work focuses on identifying, describing and offering policy solutions to some of the hardest open problems at the intersection of AI misalignment and national security: loss of control, evaluation-aware models that stand to undermine existing assessment regimes, and the handoff of the development of AI systems to AIs themselves. We are routinely requested to brief decision makers in the U.S. government's executive and legislative branch and across multiple intelligence agencies, as well as to provide input into relevant federal and state legislation and procurement efforts.

The majority of our work is non-public, including privately held demonstrations and briefing materials written for decision makers across key nodes. Our public facing output includes Assurance of Frontier AI Built for National Security, a roadmap for DoW, ODNI, NIST and CAISI to operationalize AI model reliability and governability through scheming and control evaluations within existing T&E pipelines and OT authority;
Misaligned AI as a New Insider Risk, which shows how AI models with authorized access to IL5–IL7 environments and cleared personnel should be treated according to existing insider threat programs;
The Need for Deeper, White-Box Access, on why deeper access is required to keep pace with technological progress for verifiable assurance; multiple comments on RFIs, including a comment to NIST's RFI on AI agent security; a Lawfare op-ed on keeping AI testing defense-worthy; and an analysis of NDAA provisions most likely to accelerate AI security.

This is a posting primarily intended for mid-career policy wonks interested in joining the Governance team and founding our DC office, though we are open to considering more junior applicants should they be an especially good fit for the role.

• Communicate Apollo's work and the state of AI progress, safety and security accurately and appropriately to decision makers across the USG and the intelligence community
• Proactively plan communication schedules around Apollo's technical work and releases to brief relevant audiences and decision makers
• Prepare relevant presentation material, demonstrations and act as point person to review policy submissions and briefings prepared by colleagues
• Brief and support senior leadership ahead of exchanges and meetings with relevant audiences and decision makers
• Contribute to the development and execution of our strategic vision to ensure safe and secure AI development and deployment in accordance with Apollo Research's mission and the organisation's core priorities

Requirements

• Have previously worked in a science communications heavy role, ideally communicating complex technological topics to a broad range of stakeholders at pace
• Are a…