IASEAI AMA #4: Stuart Russell & Julia Irwin (16 September 2026)

Our fourth "Ask Me Anything" webinar features IASEAI President Stuart Russell in conversation with Associate Executive Director Julia Irwin after a week of high-profile developments and growing public attention to AI safety Following his two recent op-eds, Stuart offers his thoughts on the departure and warning of Anthropic safety researcher Jacob Coxon, and why it had so much impact. He then turns to Dario Amodei's "We Must Pace the Frontier" letter, explaining his reading and arguing that the "pacing" approach is backwards: you don't just slow down to allow enough time for some safety work to happen. Safety requirements should be set first, with development proceeding only once they are met. The ensuing Q&A touches on whether the certification layer itself can be trusted when systems know they are being tested, where the burden of producing scientifically valid evidence of safety should sit, and what meaningful red lines would require. Stuart also responds to the regulatory-capture and "but China" arguments, assesses the enforceability of the EU AI Act, and makes the case for insurance and liability as forces for safety. Asked what researchers should work on now, he points to assistance games and to formal oracles. He ends closer to the present, advocating for IASEAI's Right to Know project and pointing to mounting evidence that heavy AI use erodes cognitive development in students and capability in adults. Questions were submitted in advance and raised live by attendees. Learn more and get involved at iaseai.org!