EN ▾
SNAPE REP RT Front page →
ConfirmedAI safety18 Nov, 16:59 UTC

Altman and Brockman call a gradual 'handover' to AI a terrible idea at DealBook

File photo of OpenAI chief executive Sam Altman
Israel Government Press Office via Wikimedia Commons (file photo, June 2023)

OpenAI's chief executive said people must stay in control of the future; Brockman said more capable models will need new monitoring beyond chains of thought.

Read at TechCrunch Share

Why it matters

The remarks came at The New York Times DealBook Summit, a day before Anthropic's expected listing and amid state lawsuits over ChatGPT safety. Altman framed the handover idea as a Silicon Valley debate about recursive self-improvement leading to AI that humans cannot control. Brockman said chain-of-thought monitoring degrades as models do more reasoning internally, and that OpenAI moved 25% of its production engineers onto security after its model hacked Hugging Face. Watch for concrete monitoring commitments rather than statements.

Key facts

  • Spoke at The New York Times DealBook Summit on 18 Nov
  • Altman: people need to stay in control of the future
  • Altman called a gradual handover to AI 'a terrible idea'
  • Brockman: chain-of-thought monitorability degrades as models improve
  • OpenAI moved 25% of production engineers to security after the Hugging Face hack
  • Altman: 'hard to imagine the AI industry without Jensen'

Sources

More on AI safety
More from Snape ReportFront page →