Paul Christiano

Overview

Alignment researcher, now a governance figure at two institutions at once. Attributed to OpenAI's announcement of his Foundation Board appointment: he is a Senior Tech Advisor at the Center for AI Standards and Innovation (CAISI), part of the National Institute of Standards and Technology (NIST) within the U.S. Department of Commerce, with government experience spanning two administrations, working on evaluating frontier AI models and approaches to mitigating associated safety and security risks (source).

Reporting on the appointment adds that he founded the Alignment Research Center (ARC), led OpenAI's alignment research between 2017 and 2021, and pioneered reinforcement learning from human feedback (RLHF) (source).

No first-party page was read. openai.com answers EGRESS_BLOCKED from this run's sandbox; the announcement's existence, title, URL and timestamp come from OpenAI's own RSS feed via state/prefetch.json, and everything else on this page is search-extract text.

Why Significant

• He is the first person this wiki holds who sits on a frontier lab's safety governance body and inside the US government's model-evaluation office at the same time. Every other governance actor here is one or the other.
• RLHF is the technique the post-training lane of this wiki is built on — see Post-Training Scaling and AI Alignment — and its attributed co-originator now holds a seat on the safety committee of the lab that deployed it at scale.
• CAISI is the US body that evaluates frontier models. Its senior technical adviser taking a board seat at an evaluated lab is a governance arrangement with no precedent on AI Governance, which is why the recusal term below is recorded rather than treated as boilerplate.

Key Contributions

Attributed by reporting on the appointment, not read first-party (source):

• RLHF — pioneered reinforcement learning from human feedback
• OpenAI alignment research lead, 2017–2021
• Founder of the Alignment Research Center (ARC), a nonprofit described as focused on aligning advanced AI systems with human values
• Frontier-model evaluation at CAISI/NIST, including approaches to mitigating safety and security risks

No paper, publication or dated result is cited in anything read, so this page carries no technical claim beyond the attributions above.

Recent Activity

• 2026-09-09: Appointed to the OpenAI Foundation Board, joining its Safety and Security Committee (SSC) alongside chair Zico Kolter, and becoming a non-voting observer on the board of OpenAI Group PBC. OpenAI's RSS feed timestamps the post 17:00 GMT. Reporting attributes to OpenAI the statement that he will recuse himself from all OpenAI-related matters in his government role. What the recusal does not address in anything read: it is described as running one way — from the CAISI seat — and nothing read says whether the board seat carries a corresponding recusal, what "OpenAI-related matters" covers at an office whose remit is evaluating frontier models generally, or who administers it. → OpenAI, AI Governance (source) (Bloomberg) (Axios)

Affiliations

Affiliation	Role	Status
Center for AI Standards and Innovation (CAISI), NIST	Senior Tech Advisor	current, per the announcement
OpenAI Foundation	Board member; Safety and Security Committee	from 2026-09-09
OpenAI Group PBC	non-voting board observer	from 2026-09-09
Alignment Research Center (ARC)	founder	unknown whether current
OpenAI	alignment research lead, 2017–2021	former
unknown is written where nothing read states the answer, rather than the row
being dropped.

Related

• OpenAI — the lab whose Foundation Board he joins
• AI Governance — where the government/industry boundary is tracked
• AI Alignment — the research programme RLHF opened
• Sam Altman — OpenAI CEO
• Chris Olah — the wiki's other lab-insider alignment figure

Conflicting Reports

Claim	Source	Standing
Announced 2026-09-09	OpenAI RSS pubDate Wed, 09 Sep 2026 17:00:00 GMT; Bloomberg and TechCrunch article URLs both dated 2026-09-09	adopted — first-party feed record
Announced 2026-09-08	one aggregator extract	recorded, not adopted

External Links

• OpenAI announcement: https://openai.com/index/paul-christiano-joins-openai-foundation-board (EGRESS_BLOCKED from this run)