Lama Ahmad works at OpenAI, where she has led the company's external red-teaming and third-party assessment programs. She is the named author of "Priorities and principles for effective third party assessments," published September 22, 2026, which sets out four priority areas in which OpenAI would support independent assessment of its safety work and seven principles for how such assessments should be run (Priorities and Principles for Effective Third Party Assessments (OpenAI, September 2026)). Her title is not stated in the post.
Role at OpenAI
A Data & Society event biography describes Ahmad as technical program manager of trustworthy AI at OpenAI, leading "efforts on external assessments of AI systems' impacts on society"; as part of the safety systems team, it states, she has led the researcher access program, OpenAI's red-teaming efforts, third-party assessment and auditing, and public input projects (Source: datasociety.net).
Her contributions appear in the credits of successive OpenAI model documents. The GPT-4 Technical Report (March 2023) lists her as lead of the expert red teaming and adversarial testing program. She is listed among contributors to the GPT-4o system card (August 2024) (Source: Raw Sources/GPT-4o System Card (2024-08-08).md), under External Red Teaming in the OpenAI o1 system card (December 2024), and under Safety Systems in the GPT-4.5 system card (February 2025) (Source: Raw Sources/OpenAI o1 System Card (2024-12-05).md; Source: Raw Sources/OpenAI GPT-4.5 System Card (OpenAI, 2025-02-27).md).
Publications
- External red teaming. Ahmad is lead author, with Sandhini Agarwal, Michael Lampe and Pamela Mishkin, of "OpenAI's Approach to External Red Teaming for AI Models and Systems" (arXiv 2503.16431, submitted January 24, 2025). The white paper describes the design choices in OpenAI's external red teaming — the composition of the red team, the level of model access granted, and the guidance given to testers — the outcomes it feeds, such as risk assessments and automated evaluations, and its limitations within a broader evaluation program (Source: arxiv.org). See Jailbreaking and Red Teaming.
- Interaction harms. She co-wrote, with Lujain Ibrahim, Saffron Huang, Umang Bhatt and Markus Anderljung, "Towards Interactive Evaluations for Interaction Harms in Human-AI Systems" (June 23, 2025), published in the Knight First Amendment Institute's AI essay series (Source: knightcolumbia.org).
- Third-party assessment. The September 2026 post described below.
Positions
The September 2026 post, published under her byline, states OpenAI's commitment to "supporting independent assessments with deep levels of access across training, evaluation, and deployment," proposes pre-registered safety claims as the starting point of an assessment, and states that "no one third party can or should comprehensively cover urgent frontier safety questions" (Priorities and Principles for Effective Third Party Assessments (OpenAI, September 2026)). Its four priority areas are independent assessment of safety cases, assessment of critical safeguards including misalignment and chain-of-thought monitoring, assessment of Preparedness and alignment evaluations, and independent investigation of critical misalignment incidents; its principles include proportionate access, conflict-of-interest disclosure, a remediation period before publication, and editorial independence with lab-requested redactions (Priorities and Principles for Effective Third Party Assessments (OpenAI, September 2026); see OpenAI). See Third-Party AI Assessment.
Relationships
- related: OpenAI
- related: Priorities and Principles for Effective Third Party Assessments (OpenAI, September 2026) — author.
- related: OpenAI o1 System Card (December 2024), OpenAI GPT-4.5 System Card (February 2025), GPT-4o System Card (OpenAI, August 2024) — listed contributor.
- related: GPT-4 Technical Report (OpenAI, arXiv 2303.08774, March 2023) — expert red teaming and adversarial testing program lead.
- related: Third-Party AI Assessment, Jailbreaking and Red Teaming, Knight First Amendment Institute (Knight Columbia)