Summary
On Aug 31, 2026, between approximately 7:40 a.m. and 12:53 p.m. PDT, some users experienced errors and delays when starting or continuing tasks in ChatGPT Work. Increased demand on shared network and computing resources disrupted the infrastructure supporting those tasks. Access to ChatGPT Work was restored across all subscription plans by approximately 12:53 p.m. PDT.
Impact
During this incident:
Some users were unable to start or continue ChatGPT Work tasks on web and mobile, or experienced longer response times.
Root Cause
A code change increased demand on systems supporting ChatGPT Work, while background activity generated additional network connections. Combined with normal usage, this exceeded available network capacity and disrupted connections needed to run tasks. Memory pressure and repeated retries added to the overload.
Resolution
Engineers reduced traffic to affected infrastructure, temporarily limited new tasks, and freed computing resources to relieve pressure. Our infrastructure provider added network capacity, and engineers gradually restored access as the systems stabilized.
Prevention and Improvements
We have implemented or are working on several improvements to reduce the likelihood and impact of similar incidents:
Preserve capacity for user tasks. We reduced unnecessary system activity and are improving monitoring to detect unexpected increases in demand.
Limit overload during recovery. We adjusted how failed requests are retried to reduce additional load when systems are under pressure.
Reduce network demand. We reduced unnecessary network activity and are improving how supporting systems use network connections.
Strengthen infrastructure capacity. We are working with our infrastructure provider to increase available capacity and better support periods of high demand.
We apologize for the disruption and are continuing to strengthen the reliability of the systems that support ChatGPT Work and related features.
