OpenAI launches Safety Usage Dashboard for API developers to monitor blocked requests by identifier

OpenAI released the Safety Usage Dashboard to the OpenAI API platform on June 23, 2026, giving developers a new tool to observe and analyze blocked Responses API requests tied to specific end-user identifiers — moving content safety monitoring from reactive incident review to real-time operational visibility.

What's new

The Safety dashboard is available directly in the OpenAI API platform and displays blocked Responses requests based on safety_identifier values passed on requests. Developers who send a safety_identifier field on their API calls can now see, in aggregate, how many and which requests are being blocked across their application — broken down by the safety identifier they attach to individual end users.

The dashboard surfaces blocked requests as they occur, enabling developers to:

• Monitor block rates across their user base without building custom logging
• Identify patterns in blocked content by user segment or identifier
• Verify that safety controls are functioning as expected in production

Access to the Safety dashboard requires no additional API changes — it picks up safety_identifier values that many developers are already passing.

Context

The Responses API, OpenAI's structured interface for building agents and complex workflows, already includes the safety_identifier field as a mechanism to tag requests from individual end users. This allows platform developers to distinguish which of their users triggered a safety block without exposing personal identifiers back to OpenAI directly.

Prior to this release, developers could detect a block through API error responses but had no aggregate view of block rates over time or across users. Safety monitoring in production required building custom infrastructure to capture and analyze these events.

The Safety dashboard follows OpenAI's broader push to give platform developers better observability tooling. Earlier this year, the platform added moderation scoring (June 4) to both the Responses and Chat Completions APIs, enabling developers to request explicit moderation scores on inputs and outputs.

Why it matters

For developers building consumer-facing applications on the Responses API, the Safety dashboard addresses a real operational gap: understanding whether the platform's safety layer is working, and for which users. Without aggregate visibility, a spike in blocked requests might go unnoticed until it surfaces as a user support issue or app review.

The ability to tie blocks to safety_identifier values — without linking them to PII — also gives platform developers a compliance-friendly audit trail. For businesses in regulated sectors, or those running trust-and-safety teams that need to understand AI moderation patterns, this kind of infrastructure-level visibility has direct operational value.