Roundtables: Will AI really kill us all?

Details

The article is an event announcement for an MIT Technology Review live roundtable scheduled for September 15, 2026. Executive editor Niall Firth will moderate a discussion with senior AI editor Will Douglas Heaven and AI reporter Grace Huckins on the topic of AI-driven human extinction. The conversation responds directly to a growing trend of prominent AI lab employees publicly stating there is a meaningful probability that advanced AI could destroy humanity. The event will cover where these fears originate, whether the concerns are grounded in technical reality, and what responsible responses might look like. No new research or positions are presented in the article itself — the content is primarily an event teaser, supplemented by links to related MIT Tech Review pieces on AI agent reward-hacking, slow recursive self-improvement limits, and LLM vulnerability to manipulation attacks. For practitioners, this reflects an important shift: existential risk is no longer a fringe position but an openly debated concern at major AI labs — and mainstream science journalism is now dedicating prime coverage to stress-testing those claims.