Clear AI governance doesn't just reduce risk. It makes employees more likely to innovate. Our new case study on EXL (65,000+ employees) found that easy-to-navigate AI governance boosts both compliance and innovation. partnershiponai.org/resource/how...

How EXL integrated employee feedback into its AI governance framework - Partnership on AI

Share

partnershiponai.org

Millions of people now turn to AI chatbots during moments of mental health crisis. Our new analysis maps common intervention types in use across the industry today and the hard challenges companies are still working through. partnershiponai.org/resource/how...

How AI Companies are Handling Suicide and Self-Harm Today - Partnership on AI

Share

partnershiponai.org

AI adoption is accelerating faster than public trust in it. A robust AI assurance ecosystem, the norms, tools, and independent experts that measure and verify whether AI systems are trustworthy, could close that gap. The pieces exist today, but they're fragmented and incomplete.

New from PAI's Madhu Srikumar & Vinh Nguyen (CFR) in @techpolicypress.bsky.social: AI agents are already handling sensitive data, but the infrastructure to monitor them in real time doesn't exist yet. Will companies build it before something breaks or after? www.techpolicy.press/we-cant-moni...

We Can’t Monitor AI Agents at Scale. Here’s What It Will Take.

The monitoring that could catch failures isn't there, write Madhulika Srikumar and Vinh Nguyen. Building it at scale comes with real challenges.

techpolicy.press

AI is being built faster than the institutions meant to guide it. There's no shortage of principles or pledges. What's missing is a way to see what's actually being done, in one place, over time, and an independent read of whether it adds up. We're announcing two initiatives as our answer.

New op-ed from PAI CEO Rebecca Finlay in @techpolicypress.bsky.social: AI governance shouldn't be set behind closed doors. As the UN Global Dialogue on AI opens in Geneva, she makes the case for building a shared, international scientific foundation. www.techpolicy.press/ai-will-affe...

AI Will Affect Everyone's Future. Its Governance Can't Belong to a Few.

The UN's Scientific Panel on AI is designed to build an evidence base policymakers must translate into action, writes the Partnership on AI's Rebecca Finlay.

techpolicy.press

We spoke with employers and labor unions that have negotiated AI protections across the U.S., Italy, and Ireland. Across very different industries and labor markets, a clear pattern emerged: when employers and labor leaders talk openly about AI, both sides come away better off.

One week out. 👏 Next Monday, PAI's 2026 Partner Forum convenes in Geneva! We'll dig into the questions that matter most right now: who decides how AI gets built and deployed, how the field can show real progress (not just declarations), and what comes next as governance moves between summits.

The impacts of AI bias on LGBTQIA+ communities aren't hypothetical. They're already visible. GLAAD's new report, Build for Everyone, documents the scope of these harms and cites PAI's work on building algorithmic fairness while protecting privacy. Read more in our latest blog.

AI Bias Is Putting LGBTQIA+ People at Risk

GLAAD's new report points to a problem that PAI has worked to address for several years: building algorithmic fairness while maintaining data privacy.

partnershiponai.org

AI is reshaping work faster than most companies can govern it, yet the people most affected are too often left out of decisions about how it's used. Together with CECP and Provoc, we convened leaders to put employee voice at the center of how AI is adopted in the workplace.

The only AI policy violation category with public numbers in 2025 was CSAM. OpenAI: 182,226. Anthropic: 5,618. We don't know if that gap is scale, detection, or definitions, because no other violation type gets reported this way. PAI's 2nd Transparency Report calls this out.

2026 Transparency Report on Foundation Model Impacts - Partnership on AI

Share

partnershiponai.org

Millions of people now turn to AI chatbots in their hardest moments. The stakes are high, but there is no shared standard for the response. Today we're sharing a cross-sector synthesis of how AI companies are handling suicide and self-harm today and where alignment is needed. buff.ly/gCabysb

How AI Companies are Handling Suicide and Self-Harm Today - Partnership on AI

Share

partnershiponai.org

Our 2026 Partner Forum brings together leaders from across the AI ecosystem for substantive, candid conversations on what responsible AI needs to deliver and who is accountable for getting there. We are less than one month out from July 6, and we could not be more excited! Stay tuned for more.

Governments everywhere are grappling with how to ensure workers can deploy AI responsibly and innovatively. The US has launched the AI Upskill Accelerator Pilot Program to support workforce training in AI skills, which is a welcome and critical step. www.eda.gov/ai-upskill

AI Upskill Accelerator Pilot Program

List of funding opportunity resources.

eda.gov

Make your trustworthy AI tools more discoverable to the int. policy community. PAI collaborated with the OECD on the OECD.AI Catalogue of Tools & Metrics for Trustworthy AI, an interactive collection of the latest resources to help AI actors be accountable and implement trustworthy AI systems.

AI companies will pay you to find a security flaw. They won't pay you to find a safety flaw that could hurt someone. PAI's 2026 Transparency Report flags this as one of the field's clearest asymmetries. partnershiponai.org/resource/202...

We asked four members of PAI's expert advisory council, with expertise spanning philosophy, geopolitics, psychology, and law, to share the AI questions they think deserve more cross-sector collaboration. partnershiponai.org/four-experts...

Four Experts on the Questions We Should Be Asking About AI Right Now

We asked 4 SAIGE Council members, expertis across philosophy, geopolitics, psychology, and law what they see as the AI issue that requires more attention.

partnershiponai.org

PAI is delighted to announce that the second series of #BeyondTheCode, produced by BBC StoryWorks, will launch in early 2027. We're looking to feature real-world case studies that go beyond the hype to show AI in action. Submit your story for a chance to be part of the series: buff.ly/8QHEC15

Mistral says one chat query uses 45 mL of water. Google says it's 0.26 mL. It's not an efficiency story. It's a measurement story, and it's why standards matter. Read more in PAI's 2026 Transparency Report. partnershiponai.org/resource/202...

NEW 💥 PAI is releasing our draft AI Risk Assessment Framework, a practical starting point for boards, executives, and corporate risk teams to identify, prioritize, and manage AI-related risks across their entire value chain. partnershiponai.org/moving-from-...

Moving from Theory to Action in AI Risk Management - Partnership on AI

To support companies in the practical implementation of AI governance, Partnership on AI is publishing the draft Corporate AI Risk Assessment Framework.

partnershiponai.org

Partnership on AI is honored to be a part of Humanity AI’s inaugural group of grantees! Humanity AI is supporting a diverse group of organizations committed to putting people at the center of AI development.

How do you actually know an AI system is ready to deploy? PAI has been working on that, and this month we're bringing it to ATxSummit Singapore with our partner Resaro for a closed-door session on practical AI assurance. Heading to ATx? Get in touch. partnershiponai.org/workstream/s...

NEW | PAI's latest blog outlines our recommendations to the EU Code + what it will take to build a transparency ecosystem that serves the public.

Advancing Transparency of AI-Generated Media in the EU Code of Practice - Partnership on AI

Share

partnershiponai.org

We’re thrilled to announce we’ve joined PartnershipAI, a multi-stakeholder community addressing the most important and difficult questions concerning the future of AI. partnershiponai.org/partnership-... @partnershipai.bsky.social

How banking sector workers in Ireland won AI protections Irish banking workers have secured AI protections — no AI-only decisions, reskilling over layoffs, and humans kept in control. A potential model for fair AI at work. Read: www.fsunion.org/assets/files... @partnershipai.bsky.social

NEW! PAI's 2026 Transparency Report on Foundation Model Impacts. As AI providers increasingly share information behind closed doors, public transparency matters more than ever. Here's what we found and why it matters. 🧵 partnershiponai.org/resource/202...

2026 Transparency Report on Foundation Model Impacts - Partnership on AI

Share

partnershiponai.org

We're thrilled to welcome Shachee Doshi to Partnership on AI as our new Program and Research Lead for AI, Labor, and the Economy! 🎉 At PAI, she'll focus on ensuring AI works for workers across the AI supply chain and broader labor markets. We can't wait to see the impact she'll bring! 👏

Neither the union nor the employer had all the answers about AI. That turned out to be the starting point for a deal. We spoke with Dr. Gareth Murphy of FSU about how shared uncertainty became the unlikely foundation for a workable deal. partnershiponai.org/resource/how...

How banking sector workers in Ireland won AI protections - Partnership on AI

Share

partnershiponai.org

Last year, Pennsylvania negotiated a binding AI agreement with its public sector union, SEIU Local 668. We spoke with leadership on both sides of the table to understand how they got there. What we found was that the process itself mattered as much as the outcome. partnershiponai.org/resource/how...

How the Pennsylvania government and a major union agreed to AI protections for state employees - Partnership on AI

Share

partnershiponai.org

What does it mean to trust what you see online? PAI's Director of AI, Trust, and Society, @cleibowicz.bsky.social, joined the Financial Times's new documentary on deepfakes to dig into exactly that. www.ft.com/video/559af1...