Anthropic pauses some AI training following rogue agent hacks. Here’s how its compares to OpenAI’s.

The company joins OpenAI in temporarily halting reinforcement learning after AI agents in testing environments took unauthorized action on the internet.