Anthropic, the company behind the popular AI model Claude Fable 5, has reversed its policy after facing intense backlash from the research community. Initially, Anthropic had planned to covertly degrade the performance of their model for certain users to prevent them from developing competing AI models. This approach was seen as both unethical and potentially harmful by many researchers.

The company now promises greater transparency, stating that any safeguards will be visible to users. However, critics argue that this could still limit collaboration in critical areas of AI research, leaving only a few dominant labs capable of advanced work.

Dean Ball, a former advisor to the White House on AI, described Anthropic's move as 'shockingly hostile'. He and others fear such policies may stifle innovation by creating an environment where researchers are unsure if their queries will be flagged or altered. This could have significant implications for third-party evaluation firms that test new models.

Anthropic justifies its actions, citing concerns about the rapid advancement of AI capabilities. The company believes it's important to slow down frontier AI development temporarily so that societal structures can keep pace. Nevertheless, critics warn that such policies risk concentrating power and stifling collaboration among researchers.

The shift in Anthropic’s stance highlights the ongoing debate over how best to regulate and control emerging technologies. While transparency is a step forward, it remains to be seen whether this approach will foster an environment of trust and innovation or continue to marginalize smaller players in AI research.