Shayne Longpre argues that in-house testing isn’t enough. He introduces third-party flaw disclosure: AI bug bounties, rapid-triage forms, and coordination hubs so GPAI vulnerabilities surface before they spread.

Victor Ojewale will discuss what current audit tools miss and where we have to go next. Based on 35 practitioner interviews and a scan of 400 tools, he identifies critical gaps in harms discovery & advocacy.

Cozmin Ududec, who leads the Science of Evaluations team at the UK AI Security Institute, will discuss extracting behavioral insights from large-scale AI agent evaluations that go beyond a simple pass/fail.

First up: Lucilla Sioli, the Director of the European AI Office, will explain how the EU AI Act will be put to work: from a code of practice for GPAI, funding to grow eval capacity, a Europe-wide evaluator network, and a scientific panel to advise enforcement.

⏰ Just 2 days left to submit your abstract to the Technical AI Governance workshop at #ICML2025!

📅 Deadline: 7 May 23:59 AOE (📄 full paper submission 12 May)
⚡ Fast‑track option for already peer‑reviewed work
🔗 Submit here → openreview.net/group?id=ICM...

✨ Top papers earn lightning-talk spots; all accepted work gets a poster.
🎟️ At least one author must attend in person.
❓ Questions → [email protected]
🌐 More info: www.taig-icml.com/call-for-pap...

📣 Call for Papers is Out!
www.taig-icml.com/call-for-pap...
📅 Submission Deadline: May 7
Topics of interest include: 👇

📣We’re thrilled to announce the first workshop on Technical AI Governance (TAIG) at #ICML2025 this July in Vancouver! Join us (& this stellar list of speakers) in bringing together technical & policy experts to shape the future of AI governance! www.taig-icml.com