AI leaders urge trust in industry as fears of existential risk grow

OpenAI CEO Sam Altman and Nvidia CEO Jensen Huang told a San Francisco conference that AI companies should be trusted to manage the technology's risks, with Huang arguing against new laws. Their calls come as warnings about AI's dangers intensify, including claims it could kill all humans by 2030. The debate has drawn scepticism from critics who say firms are marking their own homework.

Bottom line — Altman and Huang insist industry self-regulation is enough, but critics say firms shouldn't grade their own safety claims.

Go deeper (8)

• Altman said 'it doesn't take as much imagination as it used to' to see how AI could go wrong, but added 'we will get it right' and keep 'alignment and safety way ahead of capabilities', per BBC.
• Huang said 'we don't need new laws or regulations', describing safety as an engineering problem, and argued there should be no 'false choice' between innovation speed and safety.
• A researcher claimed AI could kill all humans by the end of the decade if unchecked; several AI executives agreed without explaining how, the Guardian reported.
• Critics argue that companies developing potentially catastrophic AI get to 'mark their own homework', since only OpenAI staff examined a recent incident, per the Guardian.
• Chinese experts see the safety debate as a 'silent AI Cold War' aimed at containing China's AI development under the guise of responsibility, according to People's Daily.
• Nick Reese, former DHS official, suggested success in AI should be measured like airline safety, where failures are rare and heavily scrutinised.
• Altman said AI firms should refrain from releasing products if they lose confidence in safety, and that competition with China is like a Cold War arms race.
• Zuckerberg argued companies have both ability and incentive to focus on alignment, citing competition and potential liability for harm.

Read the reporting