TechCongress AI Safety Fellowship

The AI Safety Fellowship will provide you with the opportunity to work with federal government AI Safety initiatives to help shape AI policy to create a more prosperous, equitable future for AI governance. Federal agencies have been tasked with many responsibilities in the 2023 Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence and could benefit from additional support to advance the safe and responsible development and utilization of AI.

This Fellowship program aims to:

1. Build and support internal capacity in federal agencies to advance AI safety consistent with the AI EO
2. Expand the number of public interest technology experts in the federal government
3. Cultivate agency experience among civil society technical experts
4. Advance a flourishing and equitable future by addressing the present risks and potential harms posed by AI

To achieve these goals, TechCongress will place several mid- to late-career technologists with significant AI technical expertise and strong socio-technical perspectives at federal agencies to support work implementing the AI executive order and other related AI safety challenges.

Responsibilities
As an AI Safety Fellow you will:

• Work directly with leadership at a federal agency for one year starting between August 2024 - September 2024, pending completion of a federal background check. Some federal agencies that have ongoing AI Safety efforts and may be options for placement include the Department of State, Department of Commerce, Department of Energy, and Department of Homeland Security.
• Perform duties to support effective execution and implementation of tasks set forth by the AI executive order. Related tasks might include:
  • Informing the automated capability evaluations other AI/ML engineers are building.
  • Communicating technical AI knowledge with stakeholders across sectors to develop guidance and best practices.
  • Advising agency leadership and agency Chief AI Officers on relevant AI policy issues.
  • Communicating complex AI concepts to both technical and non- technical audiences.
  • Informing the development of technical guidelines and best practices for AI safety.
  • Building coalitions with partners and other groups to support effective, responsible implementation of related tasks.
• Preparing AI threat assessments by working with experts across government.
• Researching relevant AI issues, including risk management, safety and security, auditing, model assessments, red-teaming, and other key topic areas.
• Support TechCongress by writing about and presenting your experience periodically, as well as representing TechCongress and the AI Safety Fellowship at meetings or events.

Eligibility
While successful candidates may not meet all of these requirements, below is what we’re looking for in AI Safety Fellows:

• Seven or more years of work experience or postgraduate study in machine learning and artificial intelligence, computer science or engineering, or other relevant technology fields.
• Tech savvy or socio-technical expertise, with experience working in or studying in the technology industry or the public interest technology sector.
• Advanced technical ability and training.
• Great interpersonal and communication skills.
• Ability to work with cross-sector stakeholders from industry, academia, civil society, government, among others.
• Ability to work in a fast-paced, collaborative environment.
• Flexibility and comfort with ambiguity in a rapidly evolving issue space.

The AI Safety Fellowship is an opportunity for mid- to late-career technology professionals to get hands-on experience working with a federal agency and support its AI policy work. The program embraces diversity and believes that a wide range of views, backgrounds, and experiences will contribute to improved policy outcomes.