OpenAI chief executive Sam Altman said people are right to be concerned about artificial intelligence as its capabilities advance rapidly, but argued that AI companies should be trusted to act responsibly.
Speaking at Salesforce’s annual conference in San Francisco on Tuesday, Altman said it had become easier to imagine ways in which AI development could go wrong. He said OpenAI and other companies understood the scale of the responsibility involved and would put safety and alignment ahead of expanding capabilities.
Altman said he was confident the industry could manage AI development safely, adding that companies should slow down or halt progress if they could not maintain that standard.
Calls for regulation and self-restraint
The comments followed renewed scrutiny after a researcher who left AI company Anthropic warned in a viral post that unchecked AI could kill all humans by the end of the decade. Several executives and experts said they agreed with the assessment, although the source said they did not explain how they reached that conclusion or how AI could bring about such an outcome.
Anthropic chief executive Dario Amodei subsequently called for all AI development to proceed more slowly and urged governments to regulate the sector. Altman, Google DeepMind co-founder Demis Hassabis and Elon Musk were among those who applauded the post.
Amodei said on Tuesday that Anthropic was discussing stronger safety standards and checks with other companies in the industry. OpenAI executive Chris Lehane said last week that OpenAI was working with Anthropic and Google DeepMind on voluntary frontier AI standards, regardless of whether governments supported the effort.
Political figures challenge industry position
Political figures in Washington expressed scepticism over leaving AI companies to police themselves. Senator Bernie Sanders said decisions about the technology should be made by the public rather than by a small group of wealthy technology leaders. Former White House strategist Steve Bannon also questioned whether major technology companies could be trusted to self-regulate.
Meta chief executive Mark Zuckerberg said AI laboratories had both the capacity and incentive to take action on safety, arguing that companies whose models caused harm could face significant liability. Nvidia chief executive Jensen Huang likewise said companies should decide whether new versions of AI systems are ready for release, describing safety as an engineering challenge and saying firms should pause if they lack confidence in their products.
Others in the field disagreed with an entirely voluntary approach. Anthropic executive Jack Clark said an unregulated AI industry would involve immense risks, while Logical Intelligence chief operating officer Patrick Hillman said technology companies had earned public scepticism and should show what development they were prepared to stop.
Source: BBC News
