Technology
Archived — This article has been archived. The information may be outdated.
Palantir seeks zero-data guarantees before using Anthropic models
NDTV··15 Sept
Large technology firms including Palantir, Nvidia and Booz Allen Hamilton may restrict advanced AI models unless Anthropic and OpenAI guarantee against misusing intellectual property, The Information reported Monday. Microsoft pitches isolated cloud environments amid the concerns. Anthropic's June Fable policy lets it retain usage logs for 30 days. CEO Dario Amodei on Saturday urged slower frontier development, backed by Sam.
Prism
What It Means For You
- Enterprise buyers may delay Anthropic or OpenAI deployments without irrevocable zero-data-retention contracts.
- Nvidia limits Anthropic use to less sensitive tasks and uses Nemotron internally, the report said.
- Booz Allen barred staff from Anthropic's commercial model on proprietary cybersecurity work.
What's Happening
- Reuters cited The Information's account of customer pushback.
- OpenAI faced scrutiny over possible training on user data for a math problem.
- Both labs say they do not train on customer data by default unless companies opt in.
Corporate AI procurement rules
- Palantir pressed Anthropic for zero-data-retention guarantees before distribution.
- Anthropic collects anonymized metadata to improve products per the report.
- Microsoft markets isolated clouds while rivals tighten model access.
all-news
