OpenAI CEO Sam Altman stated at the Salesforce Dreamforce conference that some AI incidents are inevitable. He suggested that the industry should adopt a transparent reporting culture similar to the aviation sector, citing the FAA and NTSB as models for safety and incident analysis.
Altman emphasized that AI companies must ensure safety, alignment, and monitoring efforts outpace the development of model capabilities. He noted that if these safety standards cannot be met, firms should be prepared to slow down or pause their research and development activities.
The discussion comes amid growing concerns from researchers, including former OpenAI and Anthropic staff, regarding the risks of autonomous systems. Critics argue that labs continue to develop more powerful models despite acknowledging the potential for catastrophic threats.
Meanwhile, the US government maintains a focus on technological leadership over China. David Sacks, a former White House AI advisor, recently responded to calls for development regulation from Altman and Anthropic CEO Dario Amodei by encouraging companies to proceed without seeking external permission.
