Quick answer
Anthropic has announced Enterprise Frontier Safeguards, or EFS, for eligible organizations using its most capable covered models. The planned controls combine zero data retention at Anthropic with automated misuse monitoring across a rolling window of activity stored in the customer's own cloud account. Customers can keep that data under their encryption keys, access policies, and audit logs, receive automated alerts directly, and assign any human review to their own cleared staff. EFS will roll out in phases later in fall 2026 and is not yet a generally available default. Organizations still need to define retention, access, response, and compliance policies for their own stored data.
EFS separates Anthropic's retention from customer-controlled monitoring storage
Anthropic describes Enterprise Frontier Safeguards as a way to pair zero data retention with monitoring for serious misuse. Under the planned design, activity data used for monitoring can stay in the customer's existing Amazon S3, Azure Blob Storage, or Google Cloud Storage account instead of Anthropic's infrastructure. The customer controls encryption keys, access policies, audit logging, and data management. This does not mean that no monitoring data exists: it means the customer, rather than Anthropic, can hold the activity data needed for the selected controls.
Sources: Anthropic
Automated systems inspect patterns across sessions and send flags to the customer
EFS is designed to analyze a rolling window of traffic for signals such as attempts to develop offensive cyber or biological capabilities and indications of stolen or leaked credentials. Anthropic says the automated system sends detected patterns directly to the customer. No Anthropic employee needs to perform the human review; the customer's own authorized team decides how to investigate and clear alerts. Because detection can span sessions and accounts, organizations should document the monitoring window, alert ownership, escalation path, and rules for access to the underlying activity data.
Sources: Anthropic
Storage, customer-managed keys, and automated review are separate opt-in controls
Anthropic says customer-owned storage, customer-managed encryption keys, and fully automated review can each be enabled independently. The controls do not change model behavior, API pricing, or rate limits. Anthropic does not charge for EFS, although the customer's cloud provider can bill for storage, reads, writes, and data egress. Security teams should model those operational costs and verify that key rotation, access revocation, deletion, audit export, and incident-response procedures work in the chosen cloud environment.
Sources: Anthropic
The rollout covers several Claude surfaces and cloud platforms
Anthropic plans EFS support for Claude Code, Claude Enterprise, the Claude Platform, Amazon Bedrock, Claude Platform on AWS, Google's Agent Platform, and Microsoft Foundry. The company says the rollout will happen in phases, with a goal of broader availability later in fall 2026. Eligible customers can request access. Until EFS is ready, Anthropic says qualifying customers will receive zero data retention for Fable 5 and Fable 5.1; AWS separately documents an interim offer for eligible internal workloads through December 31, 2026 on its services.
Sources: Anthropic, Anthropic, Amazon Web Services
Existing zero-retention agreements do not automatically cover every Claude product
Anthropic's current privacy guidance says an approved zero data retention agreement ordinarily applies to the Anthropic API and products using the organization's commercial API key, including Claude Code. It does not automatically apply to beta products, Workbench, Claude for Work, Claude Max, or other products unless the agreement says otherwise. EFS is a new, eligibility-based architecture with a phased rollout, so procurement and privacy teams should confirm the exact product, model, cloud route, interim period, and contract rather than inferring coverage from the announcement.
Sources: Anthropic Privacy Center, Anthropic
Customer-controlled logs still require a complete governance plan
Keeping monitoring data in a customer account can reduce provider-side retention and give the organization direct control, but it does not by itself establish compliance or prevent misuse. Before deployment, teams should define which traffic is covered, how long monitoring data remains, who can decrypt it, what an alert permits reviewers to inspect, how false positives are handled, and when legal, privacy, or security staff must be involved. They should also test whether the design meets sector-specific confidentiality, residency, records, and deletion obligations.
Sources: Anthropic, Anthropic Privacy Center
Enterprise Frontier Safeguards is not a Chat AI feature
This announcement concerns Anthropic's enterprise controls and supported cloud-partner routes. It does not add Enterprise Frontier Safeguards to Chat AI or change which models Chat AI offers. Organizations evaluating EFS should use Anthropic's access process and confirm eligibility, supported products, and contract terms directly with Anthropic or the relevant cloud provider.
Sources: Anthropic
Frequently asked questions
What readers usually ask
What is Anthropic Enterprise Frontier Safeguards?
It is a planned set of controls for eligible enterprise customers that combines zero data retention at Anthropic with automated misuse monitoring over activity data stored in customer-controlled cloud infrastructure.
Does zero data retention mean no monitoring logs are stored anywhere?
No. With customer-owned storage enabled, activity data needed for monitoring can remain in the customer's cloud account under its keys and policies rather than being stored by Anthropic.
Who reviews an Enterprise Frontier Safeguards alert?
Anthropic says automated systems send flags directly to the customer and no Anthropic human review is required. The customer's authorized personnel decide how to review and respond.
Is Enterprise Frontier Safeguards generally available now?
No. Anthropic says it will roll out in phases and aims for broader availability later in fall 2026. Eligible customers can request access.
Does Anthropic charge for Enterprise Frontier Safeguards?
Anthropic says it does not charge for EFS. A customer's cloud provider can still charge for storage, data operations, and egress.
Does EFS make a Claude deployment compliant automatically?
No. Organizations remain responsible for evaluating contracts, configuration, access, retention, alert handling, and applicable legal or regulatory requirements.
Is Enterprise Frontier Safeguards available in Chat AI?
No Chat AI integration is announced. EFS is an Anthropic enterprise control intended for specified Claude products and cloud platforms.
Evidence