ChatGPT Work and Outlook Outages Hit on August 31: What Happened and Were They Connected?

OpenAI’s ChatGPT Work and Microsoft’s Exchange Online services experienced overlapping disruptions on August 31, affecting two tools increasingly embedded in workplace routines. OpenAI reported elevated errors and latency that could prevent users from starting or continuing tasks, while Microsoft users reported problems accessing and sending email through Outlook. Downdetector reports for Outlook peaked at nearly 6,000. 31720

The incidents occurred on the same day, but the available evidence points to coincidence rather than a shared failure: neither company attributed its outage to the other, and the affected systems were described separately. 1722

What happened to ChatGPT Work?

OpenAI acknowledged the ChatGPT Work incident at approximately 11:04 a.m. ET, or 4:04 p.m. UTC. The company said users across multiple subscription plans could experience elevated errors and latency, leaving them unable to start new tasks or continue tasks already in progress. 31823

Reports suggested that Plus subscribers were particularly affected because Work mode was unavailable for some users. Downdetector had recorded more than 2,000 reports by 9:01 a.m. Pacific, according to contemporaneous coverage. 1927

That distinction matters for workplace users. A failure that blocks an existing task can interrupt a workflow already under way, rather than merely preventing someone from submitting a new question. OpenAI’s own status language described both new-task and continuing-task failures. 3

OpenAI said it had identified the issue and was implementing a mitigation. The supplied status and reporting material does not identify a specific technical root cause. Later status tracking listed the ChatGPT Work incident as resolved, although the available sources do not provide a single, consistently reported restoration time. 3429

What happened to Outlook and Exchange Online?

Reports about Microsoft’s services began around 8:30 a.m. Pacific, according to Tom’s Guide, and peaked at roughly 6,000 Downdetector reports around 9:30 a.m. Pacific. 20 Reuters separately reported that Downdetector began receiving Microsoft-related reports just before noon ET, with a peak of nearly 6,000 complaints around 12:24 p.m. ET. 17

Users reported difficulty accessing mailboxes, delays or failures when sending messages, and problems with other Outlook functions. Reuters said receiving email was the most commonly reported issue, accounting for 39% of reports in its account; other complaints involved the app and logging in. 17

Microsoft described the incident as an issue affecting access to Exchange Online and degraded performance across multiple Exchange Online features. It said the disruption was affecting customers and that it was working to restore service. 20

The outage lasted longer than the ChatGPT disruption in the contemporaneous reports available here: Outlook users were still reporting problems shortly after 4 p.m. ET. Microsoft’s public updates described ongoing remediation and testing, but the supplied sources do not establish a definitive root cause or firm all-clear time. 1724

Were the ChatGPT and Outlook outages connected?

There is no reported evidence of a causal connection. The timing was conspicuous because Microsoft and OpenAI have a long-standing strategic partnership, but the companies described different affected services and did not say that one incident caused the other. 1722

The most defensible conclusion is therefore that the outages overlapped without a disclosed common cause. Claims that the incidents reflected a coordinated failure, a partnership dispute, or a technical dependency between the two events are not supported by the sources provided.

Why the outages mattered to businesses

The incidents highlighted a broader reliability challenge for software that has become part of daily work. ChatGPT Work was not described as simply failing to answer occasional consumer questions; OpenAI said users could be unable to continue tasks already in progress. 3

Outlook, meanwhile, sits in the communication layer of many organizations. Problems receiving or sending messages, accessing mailboxes, or signing in can delay decisions and interrupt coordination even when other business applications remain available. The reported user symptoms show the practical impact, but the supplied sources do not support a broader estimate of business losses.

That makes the simultaneous outages notable without making them evidence of rivalry. Microsoft operates widely used workplace communications services, while OpenAI is developing workplace-oriented products that compete for enterprise attention. The outages show how consequential reliability is for both categories of service; they do not, on their own, demonstrate a conflict between the companies.

How this fits OpenAI’s recent outage history

The August 31 incident followed several other OpenAI service notices in the supplied record. Those included increased error rates in Workspace Agents and ChatGPT Work on web and mobile, elevated conversation errors affecting Plus, Pro, Business, and Edu users, file-upload and download problems, image-generation unavailability, and sign-up or login disruptions. 5691112

The sequence should be interpreted cautiously. The sources document multiple incidents and affected functions, but they do not establish that all of them shared a single cause or represent a measurable trend in reliability.

The same caution applies to claims about recent product-growth figures or a potential September IPO: those claims are not verified by the reliable sources provided here and should not be treated as established facts.

Bottom line

ChatGPT Work and Outlook experienced significant, overlapping service problems on August 31. OpenAI reported errors and latency that could stop users from starting or continuing work, while Microsoft’s Exchange Online disruption affected mailbox access, email delivery, sending, and related Outlook functions. Outlook reports peaked at nearly 6,000 on Downdetector. 31720

Both incidents were publicly acknowledged, but neither company supplied evidence in the provided material that the outages were connected. The clearest lesson is operational: as AI workspaces and cloud email become embedded in business processes, partial outages can interrupt ongoing workflows—not just make a service temporarily inconvenient.