The anti-piracy efforts of the RIAA and MPAA (and the publishing houses and…) are still pretty robust, but that didn’t stop any of the big AI companies from using gigatons of copyrighted material as datasets to train their LLMs. It’s why when you ask them to generate an image featuring Winnie The Pooh, they all know what you’re talking about.

But the big companies absolutely did not get permission to do this. Nor did anyone give permission to allow for the recent jailbreaks by AI task systems to hack into other companies. If that were any singular human (not of the owner class) they’d face charges under the CFAA. But since it’s a massive company with an army of blue-haired lawyers, nothing was done about it, and the companies actually bragged about their AIs escaping containment.

Extinction by AI takeover is far more interesting than extinction by global drought.