Nvidia and Booz Allen Hamilton limit Fable use as Anthropic doesn't guarantee zero data retention; source: Palantir hasn't made Fable available via its software
As paranoia rises over whether Anthropic or OpenAI could learn from their customers' intellectual property …
The Information
Context & Ripple Effects
Anthropic said in June that it would retain first- and third-party traffic on Mythos-class models for 30 days, even as it reported no universal jailbreaks in Fable 5 red-team testing. That made data handling, rather than model safety testing alone, a procurement issue for customers handling confidential material.
Nvidia and Booz Allen Hamilton limit Fable use, constraining Anthropic’s access to workflows involving their sensitive internal or customer information.
Anthropic’s lack of a zero-retention guarantee becomes an immediate adoption constraint for enterprise buyers that require stricter handling commitments.
Second-order effects
Anthropic faces pressure to make retention and data-use terms as decisive in enterprise sales as model capability, following Microsoft’s internal limits on Fable 5.
Enterprise AI teams gain leverage to route sensitive workloads toward models or deployment arrangements whose data-handling terms meet their own security policies.
Third-order effects
If comparable restrictions spread, frontier-model access will be governed increasingly as a security and data-governance decision, not simply a model-performance choice.
The divide between broadly available hosted models and deployments acceptable for confidential workloads may widen around auditable retention guarantees.
The trend: Enterprise adoption of frontier AI is shifting toward data-retention terms as a core access-control and vendor-selection criterion.
Iol deeply ironic that this whole dynamic is finally breaking into the open—all because OpenAI was racing to be first to solve a math proof. Has been clear to researchers since Anthropic's UI capabilities suddenly surged after Figma and Lovable became a heavy Claude user.
SITUATION DETECTED: Nvidia, Palantir, and Booz Allen Hamilton are restricting or eliminating the use of Anthropic and OpenAI models in some cases due to rising corporate data retention and intellectual property concerns, per The Information.
I've been saying for years that the market would discipline the excess hubris and delusions of Sam and Dario. FAFO in real-time. Steal your user's insights and you justify local, sovereign, and open weight models.
Probably unrelated to Anthropic's CEO using the “AI Safety(!!!!)” media cycle to demand regulations choking out open source competitors that let users retain their private data. Total coincidence.
some nervous Anthropic customers are restricting Fable use or demanding guarantees that Anthropic won't ever be able to retain data from their use of the model.
This will continue until there is legal clarity on data practices. Private corporate data, personal health data, patented works and more need strong protection and legal clarity (for both users and providers).