{
  "id": 7657312,
  "title": "Nvidia and Palantir Restrict Anthropic’s Fable Over Data Retention",
  "url": "https://urgent.news/2026/09/15/nvidia-and-palantir-restrict-anthropics-fable-over-data-retention",
  "topic": "ai",
  "section": "AI",
  "published": "2026-09-15T23:34:12.000Z",
  "source": {
    "name": "PYMNTS",
    "slug": "pymnts",
    "url": "https://www.pymnts.com/news/artificial-intelligence/2026/nvidia-and-palantir-restrict-anthropics-fable-over-data-retention/"
  },
  "original_language": "en",
  "account": "Nvidia, Palantir, Booz Allen Hamilton, and Novo Nordisk are imposing restrictions on where Anthropic's most advanced AI models, like Claude Fable 5, can operate, as reported by The Information. These boundaries are directly tied to Anthropic's data-retention terms. Nvidia and Booz Allen have explicit restrictions in place, while Novo Nordisk and Northrop Grumman have longer-standing controls over proprietary information. Data retention terms are becoming as crucial as model performance in determining which AI systems large enterprises deploy. Anthropic introduced a policy in June that allowed the company to retain customer usage logs for 30 days, up to two years if flagged by safety systems. This policy removed the zero-data-retention guarantee some enterprise customers had been relying on. Microsoft responded by limiting its employees' access to Anthropic's Fable while its legal team reviewed the change. This early resistance from major companies signals the challenges Anthropic faces in meeting enterprise data retention requirements. Nvidia restricts Fable to lower-stakes tasks, such as open-source software projects, relying instead on its own Nemotron models for sensitive work like supply chain monitoring. Booz Allen Hamilton has barred employees from using Fable for any work involving its proprietary cybersecurity software, citing concerns about the model potentially learning from their code. Novo Nordisk has long kept proprietary data out of any hosted Anthropic models, analyzing public documents and drafting generic material instead. Northrop Grumman, a defense contractor, runs open-source models on air-gapped servers for tasks like coding and scientific research, ensuring sensitive workloads never leave its infrastructure. Anthropic has responded to these concerns by letting enterprise customers keep required safety data on their own infrastructure under their own encryption keys, through a process called Enterprise Frontier Safeguards (EFS). However, some customers still want a stronger guarantee than EFS, with Palantir refusing to offer Fable through its platform until Anthropic provides an irrevocable zero-data-retention guarantee. The restrictions extend beyond software, as a major U.S. utility scrapped plans to test Fable on core power infrastructure after Anthropic refused to provide similar assurances. Some AI vendors, like OpenAI, are addressing the privacy-versus-safety trade-off with architectures that let safety systems analyze interactions without giving the parent company access to the underlying content. Despite these restrictions, sensitive workloads are moving towards zero-data-retention terms, customer-controlled storage, or locally run models.",
  "summary": "Nvidia, Palantir, Booz Allen Hamilton and Novo Nordisk are drawing hard boundaries around where Anthropic’s most advanced artificial intelligence models may operate, as first reported by The Information. Nvidia and Booz Allen tied their restrictions directly to Anthropic’s data-retention terms. Novo Nordisk and Northrop Grumman have longer-standing controls around proprietary information. Data…",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}