For months, regulated enterprises had a privacy problem with AI: every Claude conversation log sat in Anthropic’s own infrastructure. Banks, healthcare firms, and law firms — exactly the customers with the most to gain from AI — were also the ones most constrained by where that data could live.

Anthropic answered that directly on September 1, 2026 with Enterprise Frontier Safeguards (EFS), a new offering that keeps Claude’s safety monitoring intact while moving conversation data into cloud storage the customer controls.

What It Actually Does

Instead of Anthropic retaining logs of enterprise Claude usage, EFS routes that activity data into the customer’s own cloud account — Amazon S3, Azure Blob Storage, or Google Cloud Storage — under the customer’s own encryption keys.

Anthropic’s automated safety systems still run. The difference is that the flagged outputs go to the customer’s security team rather than requiring a human Anthropic employee to review them. The automated monitoring watches for serious misuse signals — attempts to develop offensive cyber or biological capabilities, stolen credential patterns, that sort of thing — but the data itself never leaves the customer’s environment.

Three components, all opt-in and all free:

• Customer-owned storage — activity logs land in your cloud account, not Anthropic’s
• Customer-managed encryption keys — you hold the keys, you control access
• Fully automated review — safety monitoring without human review by Anthropic staff

None of these change model behavior, API pricing, or rate limits.

Who It Covers

EFS will roll out across: Claude Code, Claude Enterprise, the Claude Platform, Amazon Bedrock, Claude Platform on AWS, Google’s Agent Platform, and Microsoft Foundry.

That breadth matters. Most large enterprises don’t use Claude through a single interface — they’re running it across coding tools, internal platforms, and multiple cloud providers simultaneously. EFS covering all those surfaces from one policy is a meaningful simplification for compliance teams.

Who Helped Build It

Anthropic says more than 100 institutions worked with them to design the architecture. Goldman Sachs, Morgan Stanley, Citi, and Visa are among those named. That’s not a press release list — it reflects the industries where the data residency problem was most acute and where AI deployment had stalled pending a solution like this.

The context matters: Anthropic received pushback from enterprise customers about its data retention policies. EFS is the direct response to that feedback, and the free pricing signals that Anthropic views this as a prerequisite for enterprise adoption rather than a premium feature.

What This Means for Business

If your organisation has been holding off on Claude deployments because of where the data lands, that barrier just got lower.

A few things to keep in mind:

For financial services and healthcare: The combination of customer-owned storage and automated (non-human) review addresses two of the most common objections from compliance teams. It doesn’t replace your own data governance policies, but it removes the “we don’t control where that data goes” problem.

For IT and security leaders: Customer-managed encryption keys mean your existing key management practices extend to Claude usage. You’re not trusting a third party’s key rotation schedule.

For legal and regulated industries: The automated-only review model means Anthropic employees aren’t reading your client interactions. That’s a meaningful shift for attorney-client privilege and similar concerns.

The rollout timeline: EFS begins rolling out in phases in fall 2026. If this is on your procurement checklist, it’s worth contacting your Anthropic account team now to understand when your configuration will be available.

The broader pattern here is worth noting. Enterprise AI adoption has largely been limited by two things: capability gaps and compliance friction. Capability gaps have largely been solved. Compliance friction — data residency, key management, audit trails — is where the next wave of enterprise deals will be won or lost. Anthropic is building for that reality.

---

Considering AI deployment for your business? Enterprise DNA’s advisory team helps organisations evaluate AI tools, build governance frameworks, and deploy responsibly. Book a discovery call to talk through your situation.