I set it down as a fact that if all men knew what each said of the other, there would not be four friends in the world. — Pascal, Pensées Holmes is referring to the recent reporting on Anthropic’s Claude Mythos Preview, an AI model it says is unusually strong at cybersecurity tasks — i.e. it can hack into almost anything. Anthropic says it is limiting access for now rather than releasing it broadly, though I think we’ve all seen enough movies to guess how that’s gonna go. Anyway, her tweet is melodramatic, but everyone instantly understands the horror it suggests: search history, cloud photos, private messages, old usernames, medical records, bookmarks, drafts, forgotten accounts — all the digital sediment of a life suddenly dredged up and dragged into public view. We have already seen lesser versions of this often enough. A stray remark from years ago is dug up and treated as a character indictment. One offensive post becomes the loose thread that unravels a person’s name, job,…
No comments yet. Log in to reply on the Fediverse. Comments will appear here.