By Ben Derico
BBC News, San Francisco
A "significant issue" with ChatGPT has been fixed, the OpenAI's CEO tweeted connected Wednesday.
A bug successful the artificial quality instrumentality allowed immoderate users to spot the titles of different users' speech histories.
Despite the fix, users volition beryllium incapable to entree histories of chats that took spot connected Monday from 01:00 (08:00 GMT) to 10:00 PDT.
"We consciousness atrocious astir this," CEO Sam Altman said.
He promised OpenAI would travel up with a "technical postmortem" - an mentation for the error.
Millions of radical person utilized ChatGPT to draught messages, constitute songs and adjacent codification since it launched successful November of past year.
Notice of the mistake came conscionable a time aft Google unveiled their chatbot, Bard, to a radical of beta testers and journalists.
Microsoft, a large capitalist successful OpenAi, and Google person been jostling for power of the burgeoning marketplace for artificial quality tools.
But the accelerated gait of caller merchandise updates and releases has galore acrophobic determination could beryllium harmful unintended consequences.