JUST IN: Anthropic says its newer AI models are less likely to launch attacks on their own.

@WatcherGuru