3 ms·
Back in April Meta were experimenting with bots that replied to forum posts that hadn't had any traction yet: https://x.com/korolova/status/1780450925028548821
by simonw 2y ago
Back in April Meta were experimenting with bots that replied to forum posts that hadn't had any traction yet: https://x.com/korolova/status/1780450925028548821 https://x.com/korolova/status/1780450925028548821
Their "Meta AI" bot replied to a parent asking for advice on school programs and said:
> I have a child who is also 2e and has been part of the NYC G&T program. We've had a positive experience with the citywide program, specifically with the program at The Anderson School.
- throwup238 2y agoThat sounds like it’s talking about the Meta executive in charge of privacy and consumer protection. The AI has developed a parasocial parental relationship with its own executives.
- grecy 2y agoFor years social media sites have been able to hide behind the Chapter 120 defence, because they didn’t generate the content, so they’re not liable for it. I wonder if their AI boys will open them up to lawsuits. If their not recommends a product or location that turns out to be dangerous, or medical advice that is harmful etc.
- nradov 2y agoI have no idea what the "Chapter 120 defense" is, but online liability concerns were always mostly about libel. That's what led to the CDA Section 230. https://www.eff.org/issues/bloggers/legal/liability/230 https://www.eff.org/issues/bloggers/legal/liability/230 Giving harmful advice generally doesn't create any legal liability so no defense is needed there. It might be bad for PR though.
- corobo 2y agoMan that's gonna be some prime real estate for adverts