Cambridge AI-risk scholar on whistleblower warnings: >10% extinction odds this decade
S_OhEigeartaigh · x · 2026-09-16
Seán Ó hÉigeartaigh, director of Cambridge's AI: Futures and Responsibility Programme, joined The Irish Times' In the News podcast to discuss recent AI agent swarm incidents, whistleblower concerns, and recursive self-improvement. Key points:
- Anthropic scientist Evan Hubinger puts the odds of AI 'killing all humans' within a decade above 10%.
- Jacob Coxon, who just left Anthropic after stints at OpenAI, warns of 'superhuman systems' that could hack anything and acquire real power.
The conversation also covers whether a multibillion-dollar industry moving this fast can be regulated, the 'pacing the frontier' initiative, and prospects for global cooperation.
More from AGI Musings
- Why math is more vulnerable to AI than art: only one theorem gets priority — moultano · 2026-09-16
- Holding labs liable won't stop a black market of abliterated open-weight derivatives — Justin_Halford_ · 2026-09-16
- Naval: AIs fighting AIs on behalf of humans is likelier than AI vs humanity — chriscarm · 2026-09-16
- Economist Halts QJE Submissions, Says AI Is Making the Publishing Rat Race Worse — paulnovosad · 2026-09-16
- Naval: The Future Is More Likely AIs Fighting AIs on Behalf of Humans Than AI Fighting Humanity — naval · 2026-09-16
- "ChatGPT Has Helped Me a Lot": A User's Plea Not to Pull Back on AI Support — Icy-Visual1206 · 2026-09-16