Anthropic Plans New Data Retention Option for Enterprise AI Customers

Anthropic is preparing a new safety system that will allow business customers to store their data on their own cloud infrastructure while meeting a 30-day retention requirement. The move addresses previous customer pushback regarding data retention policies for advanced AI models.

Artificial intelligence startup Anthropic PBC is planning to adjust its data retention policy to give business customers greater control over their information when using its advanced AI models, according to a person familiar with the matter.

Under the upcoming safety system, which is expected to roll out later year, enterprise customers will still be required to retain data for 30 days. However, users will be given the option to store that data on their own cloud computing infrastructure rather than on Anthropic's servers, the source noted on the condition of anonymity.

The development follows an announcement made by Anthropic in June, where the company stated it would retain all customer data for its more powerful Mythos and Fable models—along with future frontier offerings—for 30 days. The stated purpose of the policy was to help identify and prevent new kinds of cyberattacks. At the time, Anthropic maintained that it would not use this retained data to train its AI models.

Despite the assurance regarding model training, Anthropic acknowledged in a recent report that the policy would likely face resistance from customers accustomed to zero-retention standards, presenting potential risks to its business success if competitors did not adopt similar measures.

According to the source, work on the newer infrastructure-based system began prior to the public criticism and has involved coordination with over 100 customers operating in highly regulated industries.

Meanwhile, industry competition in data privacy features is ongoing. Longtime rival OpenAI announced that it has begun testing a method called private safety processing to avoid retaining customer data. OpenAI is currently trialing this software with organizations including Databricks Inc and major investor Microsoft Corp, with a wider release planned for September.

"Data privacy and security remain critical concerns for enterprise adoption of artificial intelligence. Anthropic's decision to allow businesses to retain data on their own cloud infrastructure while satisfying safety requirements demonstrates how AI companies must balance security protocols with customer data sovereignty. In the fast-evolving enterprise AI market, flexibility regarding data handling is essential for maintaining trust and competitiveness." — Dr. Shishir Gupta, Founder & CEO, StartupLanes