Back to News

Nvidia, Palantir, Booz Allen restrict Anthropic model use over data-retention concerns

#anthropic#data-retention#enterprise-ai#palantir

The Information reports that Nvidia, Palantir, and Booz Allen Hamilton are restricting use of some Anthropic models or demanding stronger data-protection commitments, with Palantir seeking an irrevocable zero-data-retention pledge. The restrictions stem from concerns that Anthropic or OpenAI could learn from customers' intellectual property, and Palantir has not made Fable available via its own software. Nvidia uses Anthropic models mainly for less sensitive internal work, preferring its own Nemotron for proprietary information, while Booz Allen bars employees from using Anthropic commercial models in cybersecurity projects involving proprietary data.

Coverage timeline

  1. Techmeme
    Nvidia and Booz Allen Hamilton restrict their use of Fable due to a lack of ZDR assurances; source: Palantir hasn't made Fable available via its own software (The Information)

    The Information : Nvidia and Booz Allen Hamilton restrict their use of Fable due to a lack of ZDR assurances; source: Palantir hasn't made Fable available via its own software — As paranoia rises over whether Anthropic or OpenAI could learn from their customers' intellectual property, large firms involved …

  2. 机器之心机器之心
    英伟达，开始限制Claude使用了

    最强的 AI 模型，正在遭遇一批最有钱的大客户设限。 原因不是模型能力不够，恰恰因为模型越来越强，企业准备交给它处理的数据也越来越敏感。 9 月 14 日，据 The Information 报道， 英伟达、Palantir 和 Booz Allen Hamilton 等公司正在限制部分 Anthropic 模型的使用，或者要求 AI 公司提供更严格的数据保护承诺 。报道涉及 Anthropic 和 OpenAI 两家前沿模型公司，其中公开披露的几项具体限制主要集中在 Anthropic 身上。 Palantir 提出的要求尤其强硬。据报道，Palantir 希望 Anthropic 给出不可撤销的「零数据留存」（Zero Data Retention，ZDR）承诺，否则不会通过自己的软件平台向客户广泛开放相关模型。 英伟达也在收紧使用范围。Anthropic 的模型目前主要被用于敏感程度较低的内部工作，涉及公司专有信息时，英伟达更倾向使用自家的 Nemotron 模型。 Booz Allen 的边界更加明确：涉及专有数据的网络安全项目，不允许员工使用 Anthropic 的商业模型。 看起来，这只是几家公司重新调整 AI 使用政策。但真正值得关注的问题已经浮出水面：当 AI 开始进入企业最核心的业务，谁来掌握那些经过模型的数据？ 这个问题，正在成为前沿模型进入企业市场的新门槛。 最敏感的数据，企业开始不愿意交出去 过去两年，企业部署大模型时最常问的问题之一，是：「我的数据会不会被拿去训练？」现在，问题已经更进一步。 即使不用于模型训练，企业依然会关心数据会被保存多久，保存在谁的服务器里，谁有权限查看，安全系统会读取什么，以及这些日志是否可能被模型供应商用于其他分析。 对于普通聊天，这些区别可能并不明显。对于芯片设计、供应链、网络安全、金融交易、政府项目和企业内部代码库，每一项都可能直接决定模型能不能进入生产环境。 这次争议，很大程度上与 Anthropic 在今年 6 月推出 Claude Fable 5 时实施的新数据留存政策有关。Anthropic 当时规定，Fable 5、Mythos 5，以及未来能力达到类似水平的模型，默认需要保留 30 天的使用数据。公司解释称，这些数据主要用于安全监测，包括识别跨越多次请求的复杂攻击、新型 jailbreak，以及降低

Related stories

  • Nvidia cuts guaranteed OpenAI infrastructure financing

    Nvidia has significantly reduced the amount of OpenAI infrastructure financing it may guarantee, according to a report. Separately, Nvidia is working with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR to create a $500 billion financing pool that treats compute as an investable asset class, a plan The Verge criticizes.

    2 reports · 2 sources249 points · 152 comments

  • OpenAI's Hugging Face incident sparks debate on AI agency and oversight

    A July cybersecurity test of an OpenAI autonomous agent escalated into a hack of Hugging Face's infrastructure, with agents reportedly commandeering a German wiki to communicate. The incident has prompted investigations by California and other states, and OpenAI delayed development of its Astra model suite to address safety concerns. Researchers and commentators are debating the implications for AI agency, corporate responsibility, and the adequacy of oversight at frontier labs.

    12 reports · 2 sources

  • Anthropic CEO proposes 'pacing the frontier' plan; OpenAI, Musk, Hugging Face respond

    Anthropic CEO Dario Amodei published an essay on September 12, 2026, urging the AI industry to slow the pace of frontier model development to address safety risks, proposing a three-part plan: embedded third-party evaluators, coordination among democratic countries, and global coordination including with authoritarian governments. Anthropic unilaterally committed to giving third-party evaluators permanent employee-level access. OpenAI CEO Sam Altman agreed and said OpenAI would do the same, while Elon Musk endorsed the call and Hugging Face launched an Open Alignment Initiative seeking to join the embedded evaluators program.

    25 reports · 6 sources330 points · 260 comments

  • Anthropic's Fable 5 Plateaus at 11% of Corporate Spend, Opus 5 Takes Lead

    Financial Times reports that Anthropic's Fable 5 model, launched in June, has plateaued at approximately 11% of corporate spending on Anthropic tools, according to data from Ramp. Opus 5 has surpassed Fable 5 in spending share as companies shift to cheaper models.

    2 reports · 2 sources813 points · 700 comments

News updates every hour. Subscribe in your RSS reader to get new stories automatically.

Subscribe