OpenAI chief executive Sam Altman has acknowledged growing fears about artificial intelligence, while urging the public to have confidence in AI companies to develop the technology responsibly.
Speaking at a Salesforce conference in San Francisco on Tuesday, Altman said the rapid progress of AI has made it easier to imagine serious risks.
He said people were right to be concerned, but argued that companies such as OpenAI could take steps to keep AI development under control.
Altman said AI firms would need to slow down or stop developing systems if they could no longer keep safety measures ahead of the technology’s capabilities.
His comments come as debate grows over whether governments should introduce stronger rules for the rapidly developing industry.
Some technology leaders support greater self-regulation. Nvidia chief executive Jensen Huang said at the same conference that AI companies should have a major role in deciding when new systems are ready for release. He described safety as an engineering challenge rather than a reason for new government regulations.
Meta CEO Mark Zuckerberg has also argued that AI companies have strong reasons to invest in safety. He said companies could face legal and business consequences if their systems caused harm.
However, several political figures have questioned whether the industry should be left to regulate itself.
US Senator Bernie Sanders has criticised the concentration of AI decision-making among wealthy technology executives. Former Trump adviser Steve Bannon has also argued that technology companies should not be expected to police themselves.
The debate intensified after a former Anthropic researcher warned that increasingly powerful AI systems could pose an extreme threat to humanity. Several AI experts and executives supported parts of the warning, although the claims remain disputed.
Anthropic CEO Dario Amodei has called for a slower pace of AI development and stronger government involvement in setting safety rules.
At the same time, OpenAI, Anthropic and Google DeepMind are working on voluntary industry standards aimed at improving safety checks.
The disagreement reflects a wider question facing governments and technology companies: how to encourage AI innovation while reducing the risks that could come with increasingly powerful systems.
