The Lawfare Podcast: "Scaling Laws: What Keeps OpenAI’s Product Policy Staff Up at Night? A Conversation with Brian Fuller"

Release Date: August 8, 2025
Host: The Lawfare Institute
Guest: Brian Fuller, Product Policy Leader at OpenAI

1. Introduction

In this insightful episode of Scaling Laws, a joint series by Lawfare and the University of Texas School of Law, host Alan Rosenstein delves into the intricate world of AI product policy with Brian Fuller, a leading figure at OpenAI. The conversation navigates the delicate balance between technological innovation and the establishment of robust policies to ensure AI's safe and beneficial integration into society.

2. Understanding Product Policy at OpenAI

Kevin: "AI only works if society lets it work." [02:43]

Brian Fuller elucidates the role of Product Policy at OpenAI, describing it as a strategic arm within the company that oversees safety and integrity issues. The team advises product developers on navigating policy and legal concerns while also establishing user guidelines for OpenAI's platforms, such as determining what users can and cannot ask ChatGPT.

Brian Fuller: "Product policy at OpenAI is the coolest organization... a group of people that is just uniquely intelligent but also profoundly kind." [04:23]

3. Balancing Business Interests with Policy

The discussion highlights the dual objectives of fostering AI utility and adhering to safety standards. Brian emphasizes the importance of aligning product policies with OpenAI’s long-term business goals while considering privacy and integrity.

Brian Fuller: "We need to balance business interests against privacy considerations and integrity measures to ensure users do the right thing." [06:46]

He further discusses the necessity of staying informed about the regulatory landscape to effectively strategize and implement policies that resonate with both company objectives and societal expectations.

4. Collaborating with Internal and External Stakeholders

Brian outlines OpenAI’s approach to engaging with both internal teams and external experts to formulate comprehensive AI policies. This includes hiring a select group of external policy advisors and security experts to provide specialized insights without overextending resources.

Brian Fuller: "We utilize external policy advisors and security experts to get well-rounded advice without relying on an overly large group." [09:51]

He also touches on the challenges of keeping abreast with global regulatory developments, noting the constant influx of information from various political and regulatory bodies.

5. Addressing High-Stakes AI Risks

A major focus of the conversation is the evolving landscape of AI risks. Initially concerned with toxic model responses and political biases, the discourse has shifted towards more existential threats such as the potential misuse of AI in developing bioweapons.

Brian Fuller: "What keeps me up at night is the possibility of AI aiding in the creation of bioweapons if safeguards aren't properly implemented." [31:30]

He stresses the urgency of addressing these high-stakes risks as AI models become increasingly capable of generating harmful content.

6. OpenAI’s Global Approach to Policy Making

Brian underscores OpenAI's commitment to a global perspective in policy formulation, engaging with international communities to ensure that AI benefits all of humanity.

Brian Fuller: "OpenAI's mission is to develop artificial general intelligence that benefits all of humanity... we're taking a truly global approach." [24:12]

This involves partnerships and delegations in various countries, ensuring diverse viewpoints are incorporated into policy decisions.

7. Lessons Learned: The Meta Anecdote

Brian shares a personal story from his time at Meta, highlighting the complexities of AI policy making. His attempt to simplify nudity policies led to unintended consequences, such as AI generating photorealistic nude figures, underscoring the challenges of creating effective AI guidelines.

Brian Fuller: "I learned two lessons: writing policies for AI models is hard, and it's crucial to collaborate with experts to achieve better outcomes." [44:38]

This experience emphasizes the necessity of interdisciplinary collaboration and humility in the face of complex AI behaviors.

8. Recommendations for Aspiring AI Policy Professionals

When discussing pathways into AI product policy, Brian advises aspiring professionals to cultivate critical thinking skills, preferably through a legal education, and to embrace a proactive, collaborative approach.

Brian Fuller: "There isn't a traditional path to AI product policy... law degrees are helpful for strategic thinking and critical analysis." [49:09]

He encourages individuals to engage deeply with both the technical and ethical dimensions of AI to effectively contribute to policy development.

9. Conclusion

The episode concludes with a reflection on the profound responsibilities borne by AI policy leaders. Brian Fuller exemplifies the intricate balance between fostering innovation and safeguarding societal interests, advocating for a collaborative and informed approach to AI governance.

Brian Fuller: "Setting standards and ensuring everyone is on the same page is crucial for maintaining AI safety and integrity." [35:14]

Notable Quotes:

• "AI only works if society lets it work." – Kevin [02:43]
• "Product policy at OpenAI is the coolest organization... a group of people that is just uniquely intelligent but also profoundly kind." – Brian Fuller [04:23]
• "What keeps me up at night is the possibility of AI aiding in the creation of bioweapons if safeguards aren't properly implemented." – Brian Fuller [31:30]

This episode offers a comprehensive exploration of the multifaceted challenges in AI product policy, providing valuable insights for policymakers, technologists, and enthusiasts alike.