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

The full article text is not available in MoonHub yet. You can read it at the original source.