The Production AI Institute is governed by a set of principles designed to preserve the independence and integrity of both the framework and the certification programme.
The Production Safety Framework is maintained by the PAI Framework Committee — currently being constituted from senior practitioners with demonstrated production AI deployment experience. Framework changes follow a public comment process: proposed changes are published for a minimum 30-day comment period before adoption. Any practitioner or organisation may submit comments. All comments received are published alongside the final decision.
The certification programme is governed by the PAI Certification Committee, which sets examination standards, reviews portfolios, and maintains assessment quality. CPAP and CPAA submissions are reviewed by PAI reviewers with direct production AI deployment experience. Reviewers with a commercial relationship to an applicant do not assess that applicant.
Operational decisions — exam scheduling, support, billing — are made by the PAI team. Decisions affecting the standard or assessment criteria require Committee review. No individual has unilateral authority to change the framework or revoke a certification.
Committee and Board members are required to disclose commercial relationships with AI vendors, cloud providers, and model developers. Members with relevant conflicts abstain from decisions in their area of conflict. Disclosure records are maintained and reviewed annually. PAI does not accept sponsorship from any organisation that could benefit commercially from specific framework content. See our independence policy for full details.
PAI publishes: the current version of the PSF (with version history), all public comments received on proposed framework changes, the names and affiliations of Committee and Board members, and an annual report covering certification volumes, assessment quality metrics, and framework activity. We believe transparency in governance is itself a component of institutional legitimacy.