[Critical] Issues with Box AI within Box Hubs

Final update

We recently addressed issues affecting Box. We would like to take the opportunity to further explain these issues and the steps we have taken to keep them from happening in the future. Between 2:14 PM PDT and 10:12 PM PDT on September 22nd, 2025, some users may have experienced difficulties while working in Box. During this time, users may have encountered intermittent errors while using Box AI in Box Hubs. The issue occurred as a result of a code change aimed at improving the reliability of our Box AI service. We were able to resolve the issue by reverting the problematic code change. In addition, we are improving our internal testing and validation procedures to prevent similar issues from occurring in the future. **Analysis** The code change that led to this incident unintentionally switched the underlying LLM that provides vector embeddings for Box AI in Hubs to a different model. That model did not have sufficient capacity to take production traffic and as a result, users experienced errors while interacting with Box AI in Box Hubs during this time. **Corrective Actions** Box has initiated the following corrective actions: * Improve internal testing and validation procedures

1. Postmortem · Dec 11, 08:20 UTC
   We recently addressed issues affecting Box. We would like to take the opportunity to further explain these issues and the steps we have taken to keep them from happening in the future. Between 2:14 PM PDT and 10:12 PM PDT on September 22nd, 2025, some users may have experienced difficulties while working in Box. During this time, users may have encountered intermittent errors while using Box AI in Box Hubs. The issue occurred as a result of a code change aimed at improving the reliability of our Box AI service. We were able to resolve the issue by reverting the problematic code change. In addition, we are improving our internal testing and validation procedures to prevent similar issues from occurring in the future. **Analysis** The code change that led to this incident unintentionally switched the underlying LLM that provides vector embeddings for Box AI in Hubs to a different model. That model did not have sufficient capacity to take production traffic and as a result, users experienced errors while interacting with Box AI in Box Hubs during this time. **Corrective Actions** Box has initiated the following corrective actions: * Improve internal testing and validation procedures for changes going to production * Improve observability processes for changes being deployed to production * Increase traffic in pre-production environments to exercise high traffic loads We are continuously working to improve Box and want to make sure we are delivering the best product and user exp
2. Resolved · Sep 23, 05:31 UTC
   After further monitoring, this incident is now considered resolved. Box services have been restored to full functionality. Please contact Box Support at https://support.box.com/ if you continue to experience any issues.
3. Monitoring · Sep 23, 05:16 UTC
   Our team has taken steps to remediate this issue and is seeing improvement for Box AI within Box Hubs. We are continuing to monitor for any additional impact.
4. Identified · Sep 23, 04:50 UTC
   Our team has identified the underlying cause of this issue and is working to take remediating steps. We will provide additional updates as they become available.
5. Investigating · Sep 23, 04:41 UTC
   We are investigating an ongoing issue affecting Box AI within Box Hubs. We will provide more information as soon as it is available.