"What's wrong with the AI platform?" This is a question that many AI users were asking last night.
On Tuesday (June 4th) Eastern Time, ChatGPT, a subsidiary of OpenAI, Claude, and Perplexity, a subsidiary of AnthropicAI, experienced massive crashes simultaneously, leaving global users unsure and even speculating about the mysterious power behind the collective unexpected crashes of these AI platforms.
ChatGPT experienced two crashes
The initial downtime issue occurred on ChatGPT. In the early hours of Tuesday (around 2pm Beijing time), there were many complaint posts on social platform X, and many users complained that ChatGPT's app and web version could not be logged in, including some advanced users who also encountered this problem.
Some users also complain that although they are able to log in to ChatGPT, the page will display "internal server error" after searching and querying questions, or the speed of generating answers is slow, even up to a few minutes.
Nearly an hour after the server outage issue occurred, OpenAI responded at 15:21 Beijing time that it was investigating the issue. Another hour later, at 16:32 Beijing time, OpenAI claimed to have found the problem. Half an hour later, OpenAI confirmed that the ChatGPT service had a "major malfunction" and "we are continuing to fix this issue.".
OpenAI official website announcement
At 19:45 Beijing time, OpenAI confirmed that all services were running and stated that the incident had been resolved, but did not disclose the cause or further details of the downtime.
However, less than three hours later, at 22:33 Beijing time, OpenAI announced that the ChatGPT server had experienced another outage, and the scale of this outage was even larger.
DownDetector displays the peak reports of two rounds of OpenAI service terminals on the network within the past 24 hours
It was not until 1am Beijing time on Wednesday that OpenAI released an announcement stating:
"We experienced a significant interruption that affected all users of ChatGPT. The impact included all services related to ChatGPT. The affected services did not include platform.openai.com or APIs. The incident began on June 4th at 2:15 pm Greenwich Mean Time (June 4th 22:15 Beijing Time) and was resolved on June 4th at 5:01 pm Greenwich Mean Time (June 5th at 1:01 am Beijing Time)."
Two other AI platforms have also encountered issues
At the same time as ChatGPT went down, some users complained early Wednesday Beijing time that there were also issues with Google Gemini, Claude AI, and Perplexity AI network searches under the umbrella of Aerospace. But these issues were resolved about an hour later.
According to reports, Claude's webpage reported an error at one point, while Perplexity displayed "received too many requests, server overload". It is recommended that users try again later.
Claude error page
Perplexity fault page
However, Claude resumed service after 1:10 am Beijing time on June 5th. Perplexity also confirmed around 1:30 am that its AI search tool had been restored.
It is unusual for three AI platforms to crash simultaneously
According to records on the OpenAI official website, ChatGPT has experienced three major crashes in the past 90 days (including this one). The last large-scale interruption of GPT occurred on May 23rd, with a duration of 8 hours.
The red bar represents the symbol of large-scale downtime
For this large-scale shutdown, artificial intelligence companies such as OpenAI did not provide a specific explanation for the reason. In November last year, ChatGPT also experienced a 90 minute outage, and OpenAI later revealed that the outage was related to DDoS attacks.
Previously, there were speculations that the downtime of Claude and Perplexity may not have been due to bugs or other issues, but rather due to the interruption of ChatGPT, which caused these other two AI platforms to receive too much traffic in a short period of time.
Currently, ChatGPT has approximately 180 million active users worldwide. For high-frequency users, artificial intelligence tools such as ChatGPT have become an indispensable and crucial part of their lives and work. Therefore, such a large-scale shutdown will bring a lot of trouble to their work and life.
