The AI Data Retention Tracker, release one: Anthropic's five retention states, and which cannot run zero-retention

The AI Data Retention Tracker, release one: Anthropic's five retention states, and which cannot run zero-retention

Four of the five model-and-tier cells this tracker's first release could verify from dated primary documents, every one of them an Anthropic commercial surface, retain your prompts by default. The clearest case is also the newest: Claude Fable 5.1, launched September 1, 2026 in GitHub Copilot, is one of only two Claude models in that product that cannot run unless Anthropic retains prompts and outputs, and zero-data-retention (ZDR) access exists only for enterprises approved under a time-bound exemption the changelog dates as running "through the end of the calendar year," meaning December 31, 2026 1.

The simple version of retention is a setting you switch off. Across Anthropic's own commercial surfaces it is currently three different things: a default you must override (Claude Enterprise conversations), a condition of use you accept to run the model at all (Fable 5 and Fable 5.1 in Copilot, and covered models for any organization that runs the Claude API or a cloud platform under zero retention), and a floor that cannot be lowered without the provider's own sign-off (Covered Models, held at 30 days on every platform where they are offered, with zero retention only where Anthropic expressly authorizes it). This tracker assembles those states per model and tier, with the date each policy took effect.

Four of five verified cells keep prompts unless someone acts

Method: a cell is a model-and-tier combination with a public, dated policy document from the provider itself, meaning a help center, privacy center, or changelog. A cell counts as retention-by-default if, once the model is in use, no available configuration stops prompts from being stored. Release one, as of September 2026, covers Anthropic's commercial surfaces and maps five cells, one retention state per cell (figures last re-verified October 17, 2026; this refresh adds dated entries through October 15, 2026 and moves one figure, Fable 5.1's stated retention window, from unstated to a policy-documented 30 days):

• Claude Fable 5.1 in GitHub Copilot (Pro+, Max, Business, and Enterprise plans): retention is required to operate Anthropic's safety classifiers. Copilot Business and Enterprise administrators must switch the model on themselves, and enabling it acknowledges the retention requirement. Retained data is not used to train Anthropic's models 1. Neither that changelog nor GitHub's model-hosting documentation states how long Fable 5.1's retained prompts are kept; Anthropic's Covered Models policy fills the gap platform-wide: prompts submitted to, and outputs generated by, covered models are retained for 30 days on every platform where these models are offered, then deleted automatically, except where flagged by trust-and-safety systems or where retention is legally required 123.
• Claude Fable 5 in GitHub Copilot: the same retention requirement, and the launch changelog for Fable 5 states the duration: Anthropic retains prompts and outputs for up to 30 days to operate safety classifiers, then deletes them 4.
• Every other Claude model in GitHub Copilot: operates under zero data retention 1. Claude Opus 5.5, added September 22, 2026, lands in this cell: Anthropic's launch post states "Like previous Opus models, Opus 5.5 is available with zero data retention" 5, and GitHub's model-hosting documentation keeps every Claude model except the two Fables under ZDR 2. So do the rest of the 5.5 family: Claude Sonnet 5.5, added September 28, 2026, and Claude Haiku 5.5, added October 7, 2026 67. The carve-out sits one level down, at features rather than models: the GitHub-Anthropic ZDR agreement covers generally available features, and Anthropic beta or public-preview features, such as tool search via the Messages API, are not covered and may be retained by Anthropic 2.
• Claude Enterprise conversations and projects: data is retained indefinitely by default; an owner can set a custom retention period, with a 30-day minimum counted from last activity 8. Those custom periods govern chats and projects only; the help article notes they do not reach Claude Design, Claude Tag, Claude Managed Agents, or other features that run Claude Code in the cloud 8.
• Anthropic Covered Models on a Business Associate Agreement: 30-day retention is required, and ZDR is not available unless Anthropic has expressly authorized it for the organization 9.

That is four of the five cells.

ZDR Lag: at least 121 days outside the exemption

The tracker's second metric, ZDR Lag, counts days from a model's launch to the date a typical paying customer can run it with zero retention. Fable 5.1 makes the point sharply. Eligible enterprises get ZDR on day zero: approved customers' requests route to ZDR endpoints under the exemption, and eligibility runs through the GitHub account team rather than a published setting; GitHub Support cannot determine eligibility or enable access 1. The exemption also narrows what an approved enterprise may do: customers who enable a Fable model with ZDR agree to use it only for internal operations, including developing and evaluating products for their own customers, and may not make the model endpoints or outputs available externally 2. Everyone else waits. The exemption covers Fable 5.1 and Fable 5 with zero data retention "through the end of the calendar year," in the changelog's words: December 31, 2026, on a post dated September 1, 2026. After that point, continued use of either model in Copilot requires Enterprise Frontier Safeguards (EFS), whose phase-in is still dated only as "later this fall" as of this October 17 re-verification; Anthropic's EFS announcement lists the surfaces it is built to reach: Claude Code, Claude Enterprise, the Claude Platform, Amazon Bedrock, Claude Platform on AWS, Google's Agent Platform, and Microsoft Foundry 1011. September 1 to December 31 is 121 days, so an unapproved Copilot Business organization faces ZDR ineligibility spanning at least that 121-day window, with the end point, the EFS rollout, carrying no date. Fable 5's window is longer: it reached Copilot on June 9, 2026 under the same requirement, which makes at least 205 days of ineligibility, June 9 through December 31, for an unapproved organization, a span that includes the three days the model ran before Anthropic suspended it across Copilot on June 12 and its re-enablement on July 1 41. The metric cuts both ways, and September 22 proved it: Claude Opus 5.5 arrived in Copilot with zero retention from day one, no exemption required 5. GitHub's October 15, 2026 deprecation notice makes that counterexample the reference point: Opus 4.8 and Opus 5 leave all Copilot experiences on November 16, 2026, with Opus 5.5 as the suggested alternative, automatically enabled under default model enablement unless an administrator has turned off the global default or explicitly disabled the model 12. EFS itself is pitched as "complete privacy (the same as a zero data retention policy)," with data stored in cloud infrastructure controlled entirely by the customer while automated safety monitoring continues 111.

The dated record, assembled from the provider's dated primary documents:

• October 13, 2025: the most recent systematic academic comparison we could locate, an arXiv study of ZDR architecture in Salesforce AgentForce and Microsoft Copilot 13.
• March 16, 2026: Anthropic ships custom retention controls for Enterprise, with an indefinite default and a 30-day floor 8.
• June 9, 2026: Fable 5 reaches general availability in Copilot with the retention requirement; GitHub's changelog states retained prompts and outputs are kept for up to 30 days, then deleted. The model is suspended across Copilot on June 12 and re-enabled July 1 4.
• September 1, 2026: Fable 5.1 reaches general availability in Copilot, retention attached 1. Anthropic announces Enterprise Frontier Safeguards the same day, phasing in "later this fall" 10.
• September 5, 2026: Anthropic publishes its Covered Models retention policy, effective June 9, 2026: covered-model prompts and outputs are retained for 30 days on every platform where these models are offered, and organizations running zero retention on Claude Console workspaces, Claude Code in Enterprise, Amazon Bedrock, Google Cloud's Agent Platform, or Microsoft Foundry must enable retention, workspace by workspace, to use covered models at all 3.
• September 22, 2026: Claude Opus 5.5 is added to Copilot and runs with zero data retention from day one, no exemption required 5.
• September 24, 2026: GitHub announces a global default policy for generally available Copilot features, effective October 22, 2026: unconfigured eligible features follow the administrator's chosen default, and explicit decisions are preserved 14. The Opus 5.5 changelog states that under default model enablement, new models are automatically enabled unless the global default is off or the model is explicitly disabled 15. Neither changelog says whether the Fable retention policy counts as an eligible feature under the new default, so the enablement mechanics this tracker documents stand as written on September 1; the October 22 effective date was still ahead as of this update.
• September 28, 2026: Claude Sonnet 5.5 reaches Copilot under those default-enablement mechanics and operates under ZDR with the other non-Fable Claude models 62.
• October 7, 2026: Claude Haiku 5.5 reaches Copilot the same way, under ZDR 72.
• October 15, 2026: GitHub announces that Claude Opus 4.8 and Claude Opus 5 will be deprecated across all Copilot experiences on November 16, 2026, with Opus 5.5 as the suggested alternative; Opus 4.8 remains available to individual Copilot subscribers on annual plans 12.
• End of the 2026 calendar year: the exemption expires and EFS becomes the stated condition of continued Fable use 1.

The cells where zero-retention is not purchasable

Anthropic's BAA terms now carry an exception where this tracker's September record quoted an absolute: Covered Models "require 30-day data retention and aren't available with zero data retention (ZDR) enabled unless Anthropic has expressly authorized it for your organization" 9. The same updated article adds a path September's record did not have: with the HIPAA configuration applied to Claude Code (local mode) and Claude Cowork (local mode) on a HIPAA-ready Enterprise plan, members can use Covered Models in both products under the BAA 9. The conflict this tracker documented survives in narrower form: Claude Code cannot use Covered Models while the organization has ZDR for Claude Code, again unless Anthropic has expressly authorized it 9. And the set of surfaces that qualify for neither has grown: the article's cloud-and-remote table now lists eleven features as not available while the HIPAA configuration is applied, from cloud sessions, the mobile app, and Cowork in the cloud to Code Review, Claude Security, Remote Control, Computer access, Dispatch, Routines, self-hosted environments, and session sharing links 9. A HIPAA customer's grid still has no corner that is both covered and zero-retention: chat on a Covered Model means 30-day retention, local-mode Claude Code and Cowork under the HIPAA configuration are covered but retained, and the cloud and remote surfaces are unavailable outright.

The floor is no longer a BAA-only finding. Anthropic's September 5 policy article states it for every platform: covered-model prompts and outputs are retained for 30 days "on every platform where these models are offered," a policy the article dates as effective June 9, 2026, and organizations that run zero retention on the Claude Console, on Claude Code in Enterprise, or through Amazon Bedrock, Google Cloud's Agent Platform, or Microsoft Foundry must turn retention on, workspace by workspace or subscription by subscription, to use covered models at all 3. The same article describes the guardrails around that retention: by default no Anthropic personnel can read retained conversations, human review happens only through a controlled access path recorded in a tamper-proof log, and data is deleted automatically after 30 days except where flagged by automated trust-and-safety systems or where retention is legally required 3.

Retention and training are separate rows, and the difference matters. GitHub's changelog states that data retained for Fable 5.1's safety classifiers is not used to train Anthropic's models 1. A prompt can be stored and still never become a training example, so the tracker scores the two axes independently.

The questions worth asking any vendor, in order: what is retained by default, and is ZDR purchasable for this model at all? The first answer lives in the provider's own docs, dated above. The second, for Fable 5 and 5.1 outside the exemption and for Covered Models without Anthropic's express authorization, is currently no. Opus 5.5, added September 22, is the counterexample that keeps the question honest: there the answer was yes on day one 5.