Salesforce has announced enhanced protection for API access. When the platform detects suspicious API data-access behaviour, it can take automated containment action. That can include terminating the affected session or freezing the user, with an email notification sent to an org administrator.

The current Salesforce security roadmap says rollout begins on 21 October 2026 for sandbox and production orgs. This is a recent timetable change: Salesforce previously listed 9 October for sandboxes and 16 October for production, then recorded the revised date in its 16 October change log.

This does not mean ordinary API use will routinely be blocked, and Salesforce has not published a customer-configurable list of detection thresholds. It does mean that platform and operations teams should be ready for an integration session—or the user behind it—to be contained without the usual application-side warning.

What Salesforce has confirmed

The release note describes an automated response to suspicious API data-access behaviour. Possible actions include:

  • terminating the session;
  • freezing the user; and
  • notifying the org’s administrator by email.

If Salesforce terminates a session, the user or application must authenticate again to continue. If Salesforce freezes the user, a Salesforce administrator must unfreeze that account first. The change applies to Enterprise, Performance, Unlimited and Developer editions, although Salesforce notes that it is not available in every org.

Salesforce has not said that every unusual API call will cause containment, nor published a fixed rule that customers should attempt to reproduce. The confirmed change is the platform’s ability to intervene when its detection identifies suspicious behaviour. Recommendations about inventories, alert routing, recovery tests and runbooks are therefore Ostrelis operational guidance, not additional Salesforce requirements.

Why an API security action can become an operations issue

An API identity is often attached to more than a technical connection. It may support order processing, case routing, finance extracts, customer communications or overnight synchronisation. If its session is revoked, a well-designed client may obtain a new session and continue. If the user is frozen, authentication retries alone cannot restore service.

The business impact will depend on architecture. A dedicated identity used by one integration gives investigators a relatively clear boundary. A shared account used across several applications can make the alert harder to interpret and the disruption wider. An individually named user authorising a business application can create a different problem: the service owner may not realise that a personal Salesforce identity is part of the operating model.

Containment also creates a judgement call. Restoring access too quickly could return a compromised identity to service. Waiting for an improvised investigation could extend a genuine outage. The answer is not automatic unfreezing; it is an agreed decision path supported by useful evidence.

Identify the identities behind API traffic

Start with an inventory of connected apps, external client apps, integration users and other accounts known to use APIs. Reconcile Salesforce configuration with middleware, application registers, supplier records and secret stores. For each important connection, record:

  • the application, business process and accountable owner;
  • the Salesforce user and authentication flow;
  • the objects and approximate data volumes normally accessed;
  • usual schedules, source networks and client platforms;
  • the systems that share the same identity or token state;
  • where failures are monitored and who receives them;
  • who can disable, investigate, reauthorise or restore the connection; and
  • how missed or partially completed work is reconciled.

This is also a useful reason to challenge shared integration accounts. Separate identities do not prevent containment, but they can reduce the blast radius and make activity easier to attribute. Any redesign should still follow the licences, permissions and supported authentication patterns appropriate to the organisation.

Prepare the alert and investigation route

An administrator email is only useful if it reaches an attended mailbox and the recipient knows what to do. Confirm the organisation’s Salesforce security contacts and administrative notification arrangements, then link them to the operational support route. Avoid depending on one person’s inbox, particularly outside working hours.

The first response should distinguish a familiar service interruption from possible misuse. Useful evidence can include the affected username, connected application, time, source IP, recent authentication events, objects queried, operation type and volume of data returned. Salesforce’s log-analysis guidance describes correlating login and API events with login keys, session keys, user IDs and connected-app IDs. Some of the richer event sources require Event Monitoring or other licensed capabilities, so document what is actually available in each org.

Preserve evidence before making broad changes where practical. If activity is not recognised, follow the organisation’s security incident process and consider the exposure of credentials, tokens, downstream data and related systems. If activity is legitimate, record why it differed from the expected baseline before restoring access.

Test recovery without bypassing containment

A representative non-production exercise should prove the application’s behaviour when a session is no longer valid. Confirm that it stops safely, produces a useful alert and does not enter an aggressive authentication loop. Check whether work resumes idempotently or whether a failed batch can create duplicates, gaps or out-of-sequence updates.

Separately, rehearse the administrative recovery path for a frozen test user. Confirm that authorised staff can recognise the state, review relevant activity, make an explicit restore decision and unfreeze the correct account. Salesforce exposes frozen state through the user administration experience and through the UserLogin object, where IsFrozen can support a controlled inventory or diagnostic check.

A recovery test should not attempt to trigger Salesforce’s detection logic or assume that a simulated session revocation is identical to a real security event. Its purpose is to validate the organisation’s response to the observable outcomes: lost session, frozen identity, application failure and administrative notification.

A practical readiness checklist

  1. Confirm scope. Identify production and non-production orgs, editions and important API identities rather than assuming the feature is present everywhere.
  2. Map dependencies. Connect each identity to its apps, business processes, owners, schedules and downstream consequences.
  3. Verify notifications. Make sure current administrators and security contacts route relevant alerts into an attended support process.
  4. Establish triage evidence. Document the logs, event data and supplier telemetry available without waiting for an incident.
  5. Rehearse safe failure. Validate session-expiry handling, retry limits, queue behaviour and reconciliation in a representative environment.
  6. Define restoration authority. State who investigates, who can unfreeze users and who approves return to service when activity is uncertain.
  7. Reduce shared access. Where proportionate, separate unrelated integrations and remove obsolete API permissions, sessions and credentials.
  8. Watch the rollout. From 21 October, review authentication failures, frozen-user reports, integration alerts and missed schedules more closely.

The Ostrelis view

This change is a security control, not simply another authentication deadline. The goal is to contain behaviour that may indicate compromised access. Preparation should therefore balance service restoration with investigation, rather than optimising only for the fastest possible reconnection.

The most resilient organisations will know what normal API use looks like, use identities with understandable boundaries and have a short path from Salesforce’s notification to the right technical and security owners. That discipline is useful beyond this rollout: it improves integration ownership, incident response and least-privilege decisions.

Ostrelis can help organisations map Salesforce integrations, review access and monitoring, test failure handling and build practical support runbooks. See our Salesforce managed services, data and integration support and approach and experience. Need help getting prepared? Talk to a Salesforce expert.

Need help getting prepared?

Ostrelis can help assess whether this change affects your Salesforce environment, identify dependencies and plan the work needed to prepare safely.

Talk to a Salesforce expert

Sources

  1. Core Security Requirement Updates — Salesforce Help (13 October 2026; change log updated 16 October 2026, accessed 19 October 2026)
  2. Security Enhancements May Require API Reauthentication — Salesforce Help — Release Notes (Winter ’27 release notes, accessed 19 October 2026)
  3. How to Identify Frozen Users in a Salesforce Org via SOQL — Salesforce Help (13 June 2026, accessed 19 October 2026)
  4. Salesforce Log Analysis Guide: Investigating Data Exfiltration Using Event Logs — Salesforce Help (2026, accessed 19 October 2026)