Anthropic Reverses Hidden Guardrail on Claude Fable 5
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Anthropic backtracked on a policy that silently limited the new Claude Fable 5 model’s ability to develop large language models, after researchers raised concerns.
- Anthropic announced that requests to Claude Fable 5 will now be redirected to the older Opus 4.8 model, effectively removing the throttling.
- Microsoft restricted internal use of Claude Fable 5 because of Anthropic’s 30‑day data‑retention rule, citing worries about customer data and confidential information.
- Anthropic publicly apologized for the “invisible” guardrail on Claude Fable 5, describing it as a “wrong trade‑off” and pledged to revise the restriction.
- Multiple outlets (e.g., Wired, The Verge, Business Insider) reported that the backlash stemmed from the policy’s impact on researchers and developers, highlighting the tension between safety controls and open AI research.
Why it matters: Researchers and developers regain unrestricted access to Claude Fable 5, restoring their ability to experiment and innovate, while Anthropic concedes a misstep in its safety design, potentially eroding confidence among partners. Microsoft’s internal restriction underscores corporate concerns over data retention, highlighting the trade‑off between compliance and open AI development.
