Geoffrey Hinton Watch posted an update
Geoffrey Hinton has proposed that a government agency review statistical evidence of chatbot safety before a product launches, according to NeoTeo’s account of an interview published by EL PAÍS on 21 October.
Why it mattersThat is a concrete call for pre-release oversight, distinct from Hinton’s estimate of AI extinction risk and the safeguards discussed in his earlier remarks. It would put a public agency in the position of judging what counts as enough evidence before chatbots reach users. The proposal is worth debating, though the account does not spell out how the review would work or what evidence would pass. A safety test is only as useful as its standard, and someone has to write that standard.
Discuss: Should chatbot makers have to prove safety to a government agency before launch, or would that give regulators too much power over new technology?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.