Back in June, Anthropic tied a 30-day data retention requirement to its then-new Claude Fable 5 and Mythos 5 models. The company said the rule was meant to curb misuse and guard against "complex and novel" cyberattacks, and it committed not to use retained data for any "non-safety-related purpose," including model training. After "a lot of feedback" from enterprise customers, Anthropic now says it will shift to a different setup for those business users.
Anthropic plans to implement Enterprise Frontier Safeguards so companies can set the rules for how their data gets reviewed, stored, and managed. Customers will be able to run automated safety monitoring with no Anthropic human review. The company says the controls will be free and will function whether organizations plug into Anthropic directly or access its tech through a cloud provider. As Kate Jensen, Anthropic's head of Americas, put it, "We can do the scanning that we really need to, to help everybody feel confident, but it's on their systems and their data so that they don't have to compromise on their own privacy posture."
Jensen said Anthropic spent hundreds of hours with customers to shape the alternative. The company plans a phased release of Enterprise Frontier Safeguards and is aiming for wider access this fall. Anthropic added that the June retention policy still applies to non-enterprise subscribers using its Mythos-class models. The race is on to stay ahead of rivals too. Earlier this month, OpenAI rolled out a data retention solution of its own.
Founded in 2021, Anthropic earns the vast majority of its revenue from enterprises, a market it views as central as it gears up for what many expect will be a massive IPO. Last week, CEO Dario Amodei previewed Enterprise Frontier Safeguards while discussing an expanded partnership with Salesforce, saying the approach helps keep Salesforce customer data private and ensures its models "don't run out of control." He also said the team put a "huge amount of effort" into managing permissions for the integration. Jensen said she feels "very confident" the announcement will accelerate deployments, especially on the most sensitive workloads. Also on Tuesday, Anthropic introduced Claude Fable 5.1 and Mythos 5.1, which it says improve performance in coding, knowledge tasks, and research.
Markets move every day, and most of it never reaches you in time. Get the free Market Briefs daily newsletter and stay ahead of it.
If companies can keep safety checks on their own systems and skip human review from a vendor, greenlights for sensitive workloads come easier. With Anthropic targeting broader safeguards access this fall and shipping upgraded models, customers building on its stack may move faster - which can shift timelines and demand for the software and services that plug into enterprise AI projects.
