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

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