Partnership on Artificial Intelligence to Benefit People and Society, commonly known as Partnership on AI (PAI), is a nonprofit coalition dedicated to the responsible use of artificial intelligence. Established in 2016, the organization brings together stakeholders from academia, civil society, and industry to develop best practice recommendations for the tech community. As of 2024, PAI reports 126 partner organizations across 16 countries, working on issues ranging from algorithmic transparency to the societal impacts of automation.
The coalition operates on a multi-stakeholder model, integrating perspectives from technology companies, research institutions, and advocacy groups. Its work focuses on translating principles into actionable guidance for developers, policymakers, and the public, with an emphasis on mitigating risks while fostering innovation in fields such as Machine learning and Generative AI.
History
The Partnership on AI was publicly announced on September 28, 2016, with founding members including Amazon, Facebook, Google, DeepMind, Microsoft, and IBM. Interim co-chairs were Eric Horvitz of Microsoft Research and Mustafa Suleyman of DeepMind. In January 2017, Tom Gruber, Apple's head of advanced development for Siri, joined the board. Terah Lyons became the founding executive director in October 2017, bringing experience from her role as Policy Advisor to the United States Chief Technology Officer Megan Smith.
Baidu became the first China-based firm to join in October 2018. In November 2020, PAI launched the AI Incident Database (AIID), a public tool for identifying and assessing AI-related harms. Rebecca Finlay, previously a board member, was named CEO on October 26, 2021, after serving as interim executive director. In February 2023, PAI introduced a framework for ethical synthetic media, backed by partners including Adobe, BBC, OpenAI, and TikTok, following a year-long collaborative process.
Mission and Principles
PAI organizes its work into five program areas: AI and media integrity; AI, work, and the economy; justice, transparency, and accountability; inclusive research and design; and security for AI. Each program produces methodological tools, reports, and guidelines aimed at practical impact. The AI and Media Integrity program addresses the potential for AI to amplify harmful content, while the AI, Labor, and the Economy program convenes economists and worker representatives to shape inclusive economic policies.
The Fairness, Transparency, and Accountability program develops standards for algorithmic equity and explainability, complementing the Inclusive Research and Design program, which promotes community participation in AI development. The Safety Critical AI program focuses on high-stakes sectors like medicine and transportation, overseeing initiatives such as the AI Incident Database and the SafeLife learning environment. The organization also emphasizes social utility, exploring AI applications in public welfare, education, and sustainability.
Programs and Initiatives
A key initiative is the framework for safe deployment of AI models, which guides providers in managing evolving capabilities and societal risks. PAI also collaborates with Google DeepMind on research into attrition rates among women and minoritized individuals in tech. In the area of explainability, the organization has hosted workshops on deploying explainable AI (XAI), bringing together industry experts to discuss practical challenges.
In partnership with First Draft, PAI investigated strategies for countering deceptive online content, reflecting its methodical approach to information integrity. The AI Incident Database remains a central resource, cataloging real-world cases to inform risk management practices. PAI's response to the National Institute of Standards and Technology (NIST) in August 2021 highlighted these efforts, including the ABOUT ML project for documenting machine learning lifecycles, as aligned with NIST's AI Risk Management Framework.
Impact and Governance
PAI's governance structure includes a board of directors drawn from partner organizations, with leadership transitions documented since its founding. The coalition's growth from six founding companies to over 120 partners illustrates its expanding influence across sectors. Its recommendations have been referenced in policy discussions, particularly regarding AI risk management and synthetic media governance. The organization continues to adapt its priorities as Artificial intelligence technologies evolve, balancing innovation with accountability.
As of 2024, PAI remains active in shaping global conversations on responsible AI, with ongoing projects in media integrity, labor, and safety. Its collaborative model distinguishes it from purely industry-led efforts, aiming to incorporate diverse voices into technical and policy decisions.