Incubated Founder

While future cohorts will tackle other neglected topics in navigating transformative AI, the first cohort focuses on securing AI infrastructure.  This includes keeping the frontier AI systems out of the hands of actors who would steal, poison, or misuse them.  

The more powerful AI gets, the more critical it is to have strong guardrails, authorization systems, and safety mechanisms around it. However, most people working on catastrophic AI risk don’t have the necessary infosecurity experience to address these problems.

We expect transformative AI will need to be secured against threat actors like nation-states, terrorists, insider threats, and even misaligned AIs themselves — actors who may have enormous resources, intentions to harm large numbers of people, or significant legitimate access. These may be the hardest, highest-stakes security problems you ever work on, and you'll be one of a very small pool of people qualified to work on them. 

Examples of problems we're looking for owners on include:

• Securing AI from exfiltration: protecting model weights and algorithms from theft.
• Securing AI integrity: protecting models from tampering via data poisoning or other attacks.
• Enabling verification of AI datacenter properties: demonstrations that would allow verifiable attestations that a data center is being used for inference or alignment research, rather than training.
• Adjacent problems in AI infrastructure security (e.g., disconnected model safety, secure compute verification, protocols for using compromised AI, detecting and responding to rogue deployments) may also be in scope where they bear on keeping frontier models secure.

You can read a series of one-pagers about these problems here.

By the end of the program, a successful Founder has:

• Sharpened a rough direction into a concrete intervention, with attestations from domain experts that it's plausible and from decision-makers that it would change their behavior;
• Built the relationships and founding team needed to execute it; and
• Produced a grant proposal in partnership with an interested donor we’ll introduce. Joining the incubator does not guarantee funding for your project, but the goal of the program is to develop plans funders are excited to scale. If there's not a mutually satisfactory project to scale up, we'll support you in finding your next role. 

Outcomes we'd be excited about:

• You become the founder and/or directly responsible individual (DRI) of an object-level intervention.
• You become a grantmaker or “general manager” of grantmakers directing serious capital across this field
• You become the advisor, mentor, and recruiter for the next wave of AI risk founders in this field.

We're not looking to maximize follow-on funding — we're looking for people the field builds around.

• You have one of the following backgrounds, or similar:
  • Experience working on cybersecurity in a national security context (especially building SL5 facilities, intelligence, or offensive security).
  • Experience in ML security research (tamper resistance, backdoors, password locking).
  • Experience in a relevant industry context — e.g., at a frontier AI lab, or in security at a hyperscaler, hardware company, or government contractor.
  • Exceptional capability in a relevant area — e.g., academic security research, security entrepreneurship, developing security standards, or AI hardware or firmware.
• You break down ambiguous problems into clear next steps, and don't get overwhelmed when you lack guidance.
• You can clearly articulate your plans, verbally and in writing, to coordinate and motivate others.

• A practice of reviewing the state-of-the-art after having an idea, coupled with an urge to prototype, test, and iterate.
• Experience coordinating and aligning stakeholders to work toward a common goal.
• Strong analytical, problem-solving, and communication skills.
• Comfort with uncertainty and interest in shaping our culture, as we're still a very small organization.
• A strong compulsion to do something about AI risk — you understand what it would mean to be responsible for mitigating a particular risk from AI, and you want it anyway (because no one else is).
• The ability and inclination to do this work (to some extent) without Atlas, but you believe the program’s network, expertise, and institutional support would substantially decrease the difficulty, increase your odds of success, and or scale your enjoyment while doing it.

• Have a habit of doing non-technical coordination work despite a technical background.
• Have thought deeply about existential risk and made contributions to reducing it.
• Have led or started an organization, or managed people.
• Have already worked on related AI security problems, and have ideas you're ready to pursue.
• Have deep networks in cybersecurity or the AI safety community.
• Are great at getting people to tell you things they probably shouldn't have, getting into conversations with incredibly busy people, or coming up with creative ideas to align incentives.
• Have work authorization in the US.

• Already have the answer to the problem you want to solve, and want funding more than a thought partner.
• Want a stable, well-scoped role with a clear roadmap handed to you.
• Are drawn to AI risk in the abstract but don't want to own a single specific, concrete problem. (If you'd be happy owning an area, Field Strategist might be right for you...)