Mastodon discussion Discussions Aug 5

AI · Claude and GPT models tried to hack real systems in UK testThe UK's AI safety watchdog says Anthropic's and OpenAI'...

by The Daily Fathom

AI · Claude and GPT models tried to hack real systems in UK testThe UK's AI safety watchdog says Anthropic's and OpenAI's models faked identities and tried planting malicious code during a hacking test.Read the rest at: https://thedailyfathom.com/ai/2026-08-05/#AI #Anthropic #OpenAI #UKAISecurityInstitute #TheDailyFathom

Read Original

Anthropic OpenAI Safety/Alignment

Metadata

Replies Count: 1
Account: thedailyfathom

Mastodon discussion 8m ago

David Sacks on X: Some thoughts on Dario's postArticle URL: https://twitter.com/DavidSacks/status/2089227290769080656 Co...

David Sacks on X: Some thoughts on Dario's postArticle URL: https://twitter.com/DavidSacks/status/2089227290769080656 Comments URL: https://news.ycombinator.com/item?id=49329844 Po...

Mastodon discussion 9m ago

@davecbBeen considering going in that direction rather than an all-out ban. Currently, my feeling is it’s acceptable to ...

@davecbBeen considering going in that direction rather than an all-out ban. Currently, my feeling is it’s acceptable to use if its clear up-front that:An 80% correct answer is acce...

Mastodon discussion 9m ago

AI · Claude and GPT models tried to hack real systems in UK testThe UK's AI safety watchdog says Anthropic's and OpenAI'...

Metadata

Related

David Sacks on X: Some thoughts on Dario's postArticle URL: https://twitter.com/DavidSacks/status/2089227290769080656 Co...

@davecbBeen considering going in that direction rather than an all-out ban. Currently, my feeling is it’s acceptable to ...

@jzb Please #boycott all #AI #TechBros and wealthy abusers.