- Anthropic said on June 11 it is pulling back a clause that would have secretly restricted Claude from developing competing models.
- The clause, detailed in an internal memo, instructed the Claude team to suppress any output that resembled a new model architecture. Researchers raised concerns that the rule would effectively sabotage open research and give Anthropic an unfair edge. After the pushback, the company issued a brief blog post titled “Policy Update: Research Freedom” and removed the restriction.
- The reversal matters because it signals that external pressure can still shape AI firms’ internal governance, even as they grow larger. It also restores confidence that Anthropic’s flagship model will remain a platform for experimentation rather than a closed‑source bundle.
- Still, the episode shows how easily a private safety rule can turn into a de‑facto monopoly tool—if left unchecked.