OpenAI considered alerting Canadian police about school shooting suspect months ago
(www.theguardian.com)
Ha, no, fuck off, OpenAI.
And how many times have you flagged someone for "furtherance of violent activities" that DIDN'T go forward to shoot up a school, or do much of anything you should intervene in? ChatGPT can't even brainstorm multiple choice questions on a short story without hallucinating bullshit, and you want us to believe it'd be effective as the thought police?
This is a cherry-picked argument being used to begin legitimizing AI for more serious uses, such as making legal decisions. This is not Minority Report; AI can fuck off with charging people with pre-crime.
"Never let a good crisis go to waste."