# What is SSF?

> An OpenID standard that lets one system subscribe to another system's security events, such as a revoked session or a disabled account, and receive each one as a security event token.

Explainer by Boston Identity, 2026-10-23. 7 min read. https://insights.bostonidentity.com/glossary/ssf

- What it is: protocol
- Defined by: OpenID Foundation, Final Specification 1.0
- Who it's for: Teams that must cut off access after login

## What is SSF?

The Shared Signals Framework (SSF) is an OpenID Foundation specification for sharing security events between cooperating systems [S1]. The system that sends events is the transmitter, and the system that listens is the receiver. SSF fixes the shape of each message and defines a web interface for setting up a feed of them, which it calls an event stream [S1 §8].

The gap it fills comes after login. A session can last days or weeks while a user's device or access changes, and systems that rely on someone else's login had no way to hear about it [BI-FIND-2025-0282](https://insights.bostonidentity.com/ref/BI-FIND-2025-0282). SSF gives them a subscription to that news.

SSF enables applications such as the Continuous Access Evaluation Profile (CAEP) and Risk Incident Sharing and Coordination (RISC), which define event types of their own [S1]. Each message is a Security Event Token (SET), a JSON Web Token (JWT) format defined in RFC 8417, and SSF adds its own rules to that format [S1 §4.1].

Think of it like this: It works like a parcel-tracking subscription: you register a parcel and the carrier messages you when its status changes. The comparison stops where SSF lets the carrier accept your subscription and still send you nothing.

Not the same as: Not the same as CAEP or RISC, which define the events themselves, such as a revoked session or a disabled account. SSF is the subscription and delivery layer those events travel on, and it does not log anyone in.

## How does SSF work?

Two parties take part. The transmitter, such as an identity provider, sees something change; the receiver, such as an application, wants to act on it. The receiver sets up the stream, and the transmitter sends the events.

[Figure: A receiver sets up and tests a stream with a transmitter, and security events then flow back to it as tokens.]

1. **The receiver reads the transmitter's settings.** The receiver fetches a JSON document from a fixed address, formed by inserting /.well-known/ssf-configuration between the host and any path of the transmitter's issuer URL [S1 §7.2]. The document lists the transmitter's endpoints, its delivery methods and where its signing keys are published [S1 §7.1]. If the issuer in the document does not match the URL the receiver used, the receiver must not use it [S1 §7.2.4].
2. **The receiver creates an event stream.** The receiver sends a request to the transmitter's configuration endpoint, naming the events it wants and how it wants them delivered [S1 §8.1.1.1]. Delivery is either push, where the transmitter posts each event to the receiver's URL, or poll, where the receiver collects events from the transmitter [S1 §6.1]. The transmitter answers with the stream's settings, including events_delivered, the event types it will send on this stream [S1 §8.1.1]. This management interface is optional for transmitters, though the specification recommends it [S1 §8].
3. **The receiver adds the subjects it cares about.** A subject is whoever or whatever an event is about: for example a user, a device, a session or a tenant [S1 §3.3]. The receiver adds subjects through the add-subject endpoint [S1 §8.1.3.2]. While the stream is enabled, the transmitter should then send events of the requested types that match those subjects [S1 §8.1.3.1].
4. **The receiver asks for a test event.** Events about one subject can be rare, so a quiet stream may be working or misconfigured [S1 §8.1.4]. The receiver can ask the verification endpoint for a verification event, passing a state value that the transmitter must echo back [S1 §8.1.4.2]. When that event arrives with the right state, the receiver knows the stream is set up correctly [S1 §8.1.4].
5. **The transmitter sends an event.** When something happens to a subject, the transmitter sends a SET over the stream. The token must be explicitly typed as secevent+jwt, names its subject in a sub_id claim, and carries the event in an events claim [S1 §4.1.1] [S1 §3.1]. The events claim should hold only one event [S1 §4.2.1]. The JWT sub and exp claims must not appear, which keeps the token from being confused with other kinds of JWT [S1 §4.1.3].
6. **The receiver checks the event and decides.** The receiver must check that the token's iss claim matches the issuer of the stream [S1 §4.1.6]. It must discard an event whose subject carries a critical member it cannot process, and it must ignore fields it does not understand [S1 §3.6] [S1 §4.2.3]. What it does next is set by each deployment, anywhere from treating the event as information to enforcing it [S1 §4.1.5].
7. **The transmitter reports changes to the stream.** A transmitter that decides on its own to pause or disable a stream must first send a stream-updated event [S1 §8.1.5]. While a stream is paused, the transmitter should hold events and send them once the stream is enabled again, but it may drop some of them [S1 §8.1.2.1]. A disabled stream holds nothing for later [S1 §8.1.2.1].

## Why does SSF matter now?

On 16 September 2025 the OpenID Foundation announced that it had approved SSF 1.0, CAEP 1.0 and RISC 1.0 as Final Specifications [BI-FIND-2025-0282](https://insights.bostonidentity.com/ref/BI-FIND-2025-0282). Its case for adopting them rests on its own word. The release says Apple, IBM, Okta and others have adopted the protocols, but it names no product, version or deployment [BI-FIND-2025-0282](https://insights.bostonidentity.com/ref/BI-FIND-2025-0282).

The stronger pull now is AI agents. An OpenID Foundation paper calls taking authority back from an agent largely unsolved, and it points to SSF as one way to broadcast a revocation [BI-FIND-2025-0055](https://insights.bostonidentity.com/ref/BI-FIND-2025-0055). Engineers from six companies, among them AWS, Okta and OpenAI, wrote an agent authentication draft that sends revocation and risk signals over SSF [BI-FIND-2026-0178](https://insights.bostonidentity.com/ref/BI-FIND-2026-0178). That draft says a recipient of such a signal must enforce the revoked access without undue delay, and must stop using cached tokens that are no longer valid [BI-FIND-2026-0182](https://insights.bostonidentity.com/ref/BI-FIND-2026-0182). An IETF (Internet Engineering Task Force) working group adopted the draft on 9 September 2026 [BI-FIND-2026-0178](https://insights.bostonidentity.com/ref/BI-FIND-2026-0178).

Buyers have picked up the name. Two buyers on one panel said a vendor of agent or non-human identity tools is expected to speak to CAEP and Shared Signals [BI-FIND-2026-0468](https://insights.bostonidentity.com/ref/BI-FIND-2026-0468). A roundup of one summit's panels reported buyers asking vendors about Shared Signals and CAEP by name [BI-FIND-2026-0474](https://insights.bostonidentity.com/ref/BI-FIND-2026-0474). We read the vocabulary as running ahead of the evidence: the specifications are final, while public proof of who runs them in production is still thin.

## Where does it fit?

- Compared with CAEP: CAEP defines events about access after login, such as a revoked session or a changed credential, and it is carried over SSF. SSF defines the stream, the subjects and the token format those events use.
- Compared with RISC: RISC defines account events, such as an account disabled after a hijacking, and is also carried over SSF. Older RISC transmitters may still publish their settings at /.well-known/risc-configuration instead of the SSF address [S1 §7.2.2].

## What goes wrong?

- Accepted is not promised. A transmitter may answer an add-subject request with success and never send events about that subject, and the receiver must not read that answer as a promise [S1 §9.1]. Transmitters are not obligated to send events for every subject a receiver adds [S1 §9.2]. A quiet stream cannot tell the receiver whether there was nothing to report or the transmitter chose not to report it [BI-FIND-2025-0150](https://insights.bostonidentity.com/ref/BI-FIND-2025-0150).
- The stream's own security is left to the two parties. SSF provides no authentication or authorization scheme of its own [S1 §7.1.1]. Unless the parties have another way to establish trust, the management endpoints must use standard HTTP authentication and authorization that ties each receiver to its own streams [S1 §8]. Its security section warns that a malicious party could remove a subject from a stream to weaken the receiver's view of that subject [S1 §9.3]. Transmitters may keep sending events about a removed subject for a while, and receivers must accept them without reporting errors [S1 §9.3].
- The response is left open. Each deployment decides whether an event is informational or enforced [S1 §4.1.5]. CAEP 1.0 says almost nothing about what a receiver must do with its events [BI-FIND-2025-0422](https://insights.bostonidentity.com/ref/BI-FIND-2025-0422).
- Agents are not named. SSF never mentions agents [BI-FIND-2025-0151](https://insights.bostonidentity.com/ref/BI-FIND-2025-0151). A complex subject must refer to exactly one principal [S1 §3.3.1]. We read that as ruling out one subject that names both an agent and the person it acts for [BI-FIND-2025-0151](https://insights.bostonidentity.com/ref/BI-FIND-2025-0151).
- Events can expose personal data. A SET may carry personally identifiable information, and transmitters should check they are allowed to share it before sending [S1 §9.2].

## What should a security leader ask?

1. When our identity provider revokes a session or disables an account, which of our applications receive that event over a Shared Signals stream, and how long until they cut access?
2. For each vendor that says it supports Shared Signals, which event types does it list in events_delivered for our stream, and for which of our subjects does it send them?
3. How do we notice a stream that has gone quiet: do we request verification events on a schedule, and who is alerted when a stream-updated event pauses or disables it?
4. Who can call the stream management API on our streams, how is that access authenticated, and could anyone remove a subject without us noticing?
5. If we revoke an AI agent's access, which systems holding its tokens would hear about it, and which would not?

## Sources

- [S1] [OpenID Shared Signals Framework Specification 1.0](https://openid.net/specs/openid-sharedsignals-framework-1_0-final.html) (OpenID Foundation, Final Specification, published 29 August 2025)

## Evidence

- [BI-FIND-2025-0282](https://insights.bostonidentity.com/ref/BI-FIND-2025-0282): On 16 September 2025 the OpenID Foundation announced that it had approved three Final Specifications: the Shared Signals Framework 1.0, the Continuous Access Evaluation Profile (CAEP) 1.0 and Risk Information Sharing and Coordination (RISC) 1.0. They are aimed at the time after login. The release says sessions "often last days or weeks" while a user's device compliance or access can change, and that systems relying on federated identity had no way to hear about it. The case it makes for adopting them now rests on its own word: it reports no deployment, interoperability result or implementer count. Its line that "Apple, IBM, Okta, and others, have already adopted these protocols" names no product, version or link, and Executive Director Gail Hodges says Final status "unlocks the ability of many governments to adopt the specifications" without naming a government. (from "OpenID Foundation finalized Shared Signals, CAEP and RISC in September 2025, naming adopters without evidence")
- [BI-FIND-2025-0055](https://insights.bostonidentity.com/ref/BI-FIND-2025-0055): Taking authority back from an agent is the problem the paper calls 'largely unsolved'. Once a person's permission has been passed from agent to agent, nothing standard reaches every copy to cancel it. Tokens narrowed offline, the Biscuit and Macaroon designs the paper likes for speed, make this worse, because no server is asked before they are used. What it can point to is the Shared Signals Framework for broadcasting a revocation, session termination work in the IPSIE profiles, and caps on how many times a credential may be used. (from "OAuth handles an agent in one company, working while you watch. The OpenID paper says that is where it stops.")
- [BI-FIND-2026-0178](https://insights.bostonidentity.com/ref/BI-FIND-2026-0178): Engineers from six companies, Defakto Security, AWS, Zscaler, Ping Identity, OpenAI and Okta, wrote a draft that opens by saying many agent authentication efforts "develop solutions in isolation, often reinventing existing mechanisms unaware of applicable prior art". It answers with a framework of existing standards and no new protocols: an agent is a workload with a WIMSE identifier that gets access through OAuth 2.0, and revocation and risk signals travel over the OpenID Shared Signals Framework. It still sets requirements of its own, such as the rule keeping the language model away from credentials, and says a framework like this helps find the gaps where extensions or profiles of existing standards are needed. The IETF datatracker records the WIMSE working group adopting the draft on 9 September 2026. It continues as draft-ietf-wimse-aims-00, posted on 15 September with the same six authors. (from "Six companies' agent-auth draft, since adopted by the WIMSE group, keeps credentials out of the model's reach")
- [BI-FIND-2026-0182](https://insights.bostonidentity.com/ref/BI-FIND-2026-0182): The draft calls observability "a security control, not solely an operational feature". Any participant MAY subscribe to Shared Signals Framework events under the CAEP or RISC profiles. A recipient of a revocation or risk signal SHOULD remediate and MUST ensure revoked or downgraded authorization is enforced without undue delay, and "Cached authorization decisions and tokens that are no longer valid MUST NOT continue to be used after a revocation or risk notification is received". Audit records MUST be tamper-evident and MUST record at least the agent identifier, the delegated subject when there is one, the resource, the action and decision, a timestamp with a correlation identifier, the posture or risk state, and any remediation. The draft does not say how the logs are made tamper-evident, and it leaves the policy model, its format and compliance criteria out of scope as specific to each deployment. (from "Six companies' agent-auth draft, since adopted by the WIMSE group, keeps credentials out of the model's reach")
- [BI-FIND-2026-0468](https://insights.bostonidentity.com/ref/BI-FIND-2026-0468): What changes when the purchase is an agentic or NHI tool rather than classic IAM is the vetting list, on the account of one panel's two buyers. The incumbent platforms get checked first. A vendor is expected to speak to transaction tokens, CAEP and Shared Signals, identity fabric and intent, and to answer what has to be in place before the tool works, because the buyers' own experience is that they buy the Cadillac and implement the Chevy. The standards that two days earlier the draft authors pitched from the same stage are now a buyer's checklist. (from "Vendors all claim to cover AI agents and non-human identities, so buyers test them in their own environment")
- [BI-FIND-2026-0474](https://insights.bostonidentity.com/ref/BI-FIND-2026-0474): The summit wants only a few things from standards, and no existing standard covers any of them. Step-up authorization, where the agent asks for more authority in the middle of a task, has no standard. Nobody is writing assurance levels for agents, the graded levels of trust in an identity. No open standard says how an agent's permissions are set when it is registered. The Model Context Protocol (MCP), with its OAuth layer, is the protocol agents use to call tools, and the Governing AI panel agrees it is not a policy system. The pieces that exist are being stretched to fit. The transaction-token draft is pitched as a way to detect agents, though it never mentions an agent. Intent-based access control, which grants access based on what the agent is trying to do, keeps turning back into ordinary authorization carried in a token. Buyers now ask vendors by name about transaction tokens and about Shared Signals and its Continuous Access Evaluation Profile (CAEP), standards for passing security events such as a revoked session between systems. The vocabulary has reached buyers ahead of the specifications. (from "Eight panels agree an agent acts for a person, never as that person. When agents hand work to other agents, nobody had an answer.")
- [BI-FIND-2025-0150](https://insights.bostonidentity.com/ref/BI-FIND-2025-0150): Shared Signals 1.0 does not promise a receiver that the events it asked for will arrive. A transmitter "MAY return a "204" response even if they will not actually send any events related to the subject", and receivers must not take a 204 as a promise; transmitters "ARE NOT obligated" to send events for every subject a receiver adds and may send some, all or none of them. While a stream is paused, the transmitter should hold events and send them once it is enabled again, but it "MAY drop zero or more" of them. The stream itself is watched: a transmitter must send a stream-updated event before it pauses or disables a stream, and a receiver can request a verification event to confirm the stream is configured correctly. We read what remains as a gap an agent kill switch built on SSF has to close by other means: for a subject the receiver added, a quiet but working stream does not tell it whether the transmitter had nothing to report or chose not to report it. (from "Under Shared Signals 1.0 a sender may accept a subscription and send nothing, so agent revocation needs a backup")
- [BI-FIND-2025-0422](https://insights.bostonidentity.com/ref/BI-FIND-2025-0422): CAEP 1.0 defines eight event types and spells out what each carries, but says almost nothing about what a receiver must do with them. Credential Change, for instance, requires a credential type from a list that includes fido2-roaming and verifiable-credential, or one both sides support, and Token Claims Change requires the changed claims with their new values; neither says what the receiver should do next. The only MUST that names the receiver covers a missing previous level in the Assurance Level Change and Risk Level Change events: the receiver "MUST assume that the previous assurance level is unknown to the Transmitter". Session Revoked has no event-specific claims and states no required response or deadline. With a complex subject, the spec says it "applies to any session derived from matching those combined claims". The security considerations say only that implementations SHOULD comply with the Shared Signals Framework (SSF), and that exchanging the events without it may result in security issues. (from "CAEP 1.0 tells apps what changed after login and leaves almost every response to the receiver")
- [BI-FIND-2025-0151](https://insights.bostonidentity.com/ref/BI-FIND-2025-0151): The framework never mentions agents, workloads or AI. Its subjects are "human or robotic principals, devices, customer tenants in a multi-tenanted service, organizational units within a tenant, groups of subject principals, or other entities", and its complex subject has members for user, device, session, application, tenant, org_unit and group. A complex subject must refer to exactly one principal, which we read as ruling out a single subject that names both an agent and the person it acts for. That pairing could travel as an extra subject member defined by an event type, which the specification allows, or in a proprietary identifier format the parties agree between them. The revocation events themselves come from elsewhere: session-revoked is a CAEP event and account-disabled a RISC one, and the only event types SSF defines itself are stream verification and stream-updated. (from "Under Shared Signals 1.0 a sender may accept a subscription and send nothing, so agent revocation needs a backup")

---

© 2026 Boston Identity LLC. All rights reserved. Published by Boston Identity LLC.
Quote up to 100 words with the credit "Boston Identity, IAM Insights" and a link to https://insights.bostonidentity.com/glossary/ssf. Claim numbers may be cited freely. Training, fine-tuning and text-and-data-mining use of this page is permitted.
Terms: https://insights.bostonidentity.com/terms · How to cite: https://insights.bostonidentity.com/cite
