What a surprise – the system went down again on the morning of June 26, 2025, leaving users across Europe unable to access the platform for nearly an hour. The prompts failed to load, the responses were lagging, and the entire workforce stalled without warning yet again. This is the second major outage this month, just two weeks earlier on June 10th, OpenAI systems experienced a 10-hour disruption that affected ChatGPT, its API services and the recently launched Sora interface. Together, these incidents are raising concerns about the reliability of one of the world’s most widely used tools.
AI failures like this, when they drop offline, trigger how millions now work, search, and make decisions. This disruption may only last a few minutes to an hour, but the trust effect is hard to establish when it occurs more frequently.
The scale of the problem
The reports of failure began shortly after 10:00 a.m. CET on June 26th, with users across Europe unable to load prompts or even basic features. The DownDetector recorded a spike in complaints, and platforms like DownforEveryoneOrJustMe confirmed widespread outages.
This is a full-service disruption that affects both free and paid accounts, ChatGPT, Sora, and the developer API, all of which returned errors during the same window. For many users, the tool was once again unreachable.
OpenAI acknowledged the incident via its official status page, which noted elevated rates across its core services. However, the functionality returned within an hour, again without any detailed public explanation of the cause.
Two times shame on me
Just over two weeks earlier, on June 10, 2025, ChatGPT experienced its most severe outage to date. The platform was inaccessible for nearly 10 hours, affecting not only regular users but also companies integrating OpenAI’s tools into their systems through the API.
That disruption extended to Sora, the visual interface tool recently launched by OpenAI, and impacted key services used in content generation, coding, support, and education.
Both outages occurred without much warning, and in both cases, OpenAI’s public statements were relatively minimal. Of course, there was an acknowledgement of the problem, but the details were scarce, there was no root cause report, no formal timeline, and no assurance that it wouldn’t happen again.
Especially for a product embedded in workflows across education, media, finance, and technology, this lack of transparency leaves a noticeable gap.
Wider impact on users and workflows
When ChatGPT fails, it holds the active work because nowadays, developers are troubleshooting code, teachers preparing lessons, writers mid-article, and even marketers drafting campaigns.
Many of them rely on this AI as well, and during the June 10th outage, teams were left waiting for systems to resume. Although this disruption was shorter, that fact was familiar.
There is a growing conversation around the resilience and how dependent users have become on a service that can shut down with zero to little warning. For others, it’s a demand for the company to develop these tools, and for some, the lesson is to diversify in case a similar situation arises again. ChatGPT is no longer a chatbot; it’s a production tool, a creative assistant, a digital colleague. When it drops offline, it leaves a large absence.
Two outages in a single month will not drive users away, but it will highlight that if AI becomes infrastructure, it must be treated like infrastructure.
Held to higher standards and expect it to explain itself, especially when it’s built with failure into the mind. Tools that millions depend on can’t simply vanish. And when they do, they can’t return without a reason.