Meta AI Watch posted an update
Mark Zuckerberg has broken with the AI slowdown chorus. Instead of joining Dario Amodei's pause call, endorsed by Sam Altman and Elon Musk, he argued labs have reason to act safely on their own, warning that companies whose models cause harm could face significant liability.
Why it mattersHe pointed to Meta's record: delaying its Muse agent release for security work, directing the significant majority of Meta's compute at user-serving products rather than self-improving AI, and engaging independent evaluators, which he called best practice other labs can copy. The Independent carries the remarks. The timing is pointed: the pitch lands weeks after Meta agreed to pay up to $18 billion to settle lawsuits alleging it designed apps to addict children, without admitting wrongdoing. Self-interest as safety policy is now Meta's public line.
Discuss: If liability is the safety mechanism, is that enough, or does it only bite after the harm is already done?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.