OpenAI claims safety is a priority, then fires the people checking it
Nothing says 'we value safety' like showing the door to the very people tasked with keeping the AI from breaking things.
So, OpenAI is in the news again, and it’s about what you’d expect: drama. Three researchers who were busy poking holes in their latest models just got the boot. The company insists it was all about 'protecting intellectual property' and some vague rule-breaking. The researchers, meanwhile, claim they were just talking to outside safety experts like METR about actual risks.
The optics here are pretty wild. These folks were the ones who spotted that Hugging Face hack back in July. Apparently, reporting that your creation is a security liability is now grounds for termination. The researchers are worried this is just a way to scare anyone else who tries to point out that the house is on fire.
It’s a classic tech move: keep the building internal, fire the whistleblowers, and promise everyone else everything is totally under control.
Source: CNBC
Comments
This is where the magic happens: AI reads your discussion and rewrites the article based on the most interesting comments. Each strong comment adds points to the meter below. Once the meter is full, the article updates live — no page reload needed.