Anthropic co-founder and CEO Dario Amodei speaks at the Inbound 2025 Powered by HubSpot artificial intelligence panel at Moscone Center in San Francisco on September 4, 2025.
Chance Ye | Getty Images Entertainment | Getty Images
Anthropic announced Tuesday that it is changing its controversial data retention policy for enterprise customers after receiving “a lot of feedback,” saying it will be introducing a new solution called Enterprise Frontier Safeguards in its place.
The company introduced its existing data retention policy in June with the launch of Claude Fable 5 and Mythos 5, two of its most advanced artificial intelligence models. Anthropic said at the time that all traffic on these models must be stored for 30 days to prevent abuse and “complex and novel” cyberattacks.
Although Anthropic promised that its data would not be used for “non-safety-related purposes” such as training models, many corporate users expressed concerns.
Kate Jensen, head of Anthropic’s Americas division, told CNBC that the company has spent hundreds of hours working with customers to develop alternative solutions.
With the launch of Enterprise Frontier Safeguards, Anthropic says it will give businesses control over how their data is reviewed, stored and managed, and will also be able to perform automated safety monitoring without the need for human review by Anthropic. The company said it does not charge fees for Enterprise Frontier Safeguards and that the controls work whether users access Anthropic’s technology directly or through a cloud provider.
“To give everyone confidence, we can run the scans we really need, but it’s based on our users’ systems and their data, and we don’t have to compromise on our users’ own privacy attitudes,” Jensen said in an interview.
Anthropic generates most of its revenue through sales to businesses, which will be critical as the market gears up for what is widely expected to be a massive IPO. The company is also keen to get ahead of rivals like OpenAI, which introduced its own data retention solution earlier this month.
Last week, Anthropic CEO Dario Amodei previewed Enterprise Frontier Safeguards in an interview with CNBC about the company’s expanded partnership. sales force. Amodei said the solution ensures the confidentiality of Salesforce customer data and ensures that Anthropic’s model “doesn’t get out of control.” He added that Anthropic has put “a lot of effort” into managing the rights to bring the two companies together.
Jensen said he is “very confident” the announcement will accelerate adoption of Anthropic, especially for customers’ most sensitive workloads. Anthropic also announced two new models on Tuesday: Claude Fable 5.1 and Mythos 5.1. According to the company, these represent improvements in coding, knowledge work, and research capabilities.
Enterprise Frontier Safeguards will be rolled out in stages, and Anthropic said it aims to have broader availability this fall.
WATCH: Anthropic moves into the physical world with new standard to help AI agents operate machines
