Summary

  • Dario Amodei’s contingent at OpenAI was growing concerned about the behaviour of Sam Altman, OpenAI’s CEO.
  • They were stunned to discover the promises Altman made to Microsoft about what technologies it would get access to and were concerned these commitments would impede AI safety.
  • They also grew more paranoid about issues such as the possibility of a powerful misaligned system and what would happen if different companies caught on to OpenAI’s secrets.
  • There was also much unease about Altman’s analogy between OpenAI and the Manhattan Project, and whether the company was building the equivalent of a nuclear weapon.
  • At the same time, Altman was himself paranoid about people leaking information and also commissioned a countersurveillance audit to check for bugs Elon Musk may have left to spy on OpenAI.

By Karen Hao

Original Article