A Small AI Apocalypse Last Night: Millions Panic as They Lose Their 'Brains'
The professional world was struck by what felt like a small apocalypse. The world’s largest AI services, ranging from OpenAI’s ChatGPT and Anthropic’s Claude to Google’s Gemini and X’s Grok, suddenly ceased to function simultaneously. This event took place on Thursday morning US time, which corresponds to Thursday evening through early Friday morning WIB.
According to information reported to Downdetector, disruptions were recorded simultaneously between approximately 05:30 and 06:00 US Eastern Time, or 16:30 to 17:00 WIB on the same day. Users in Indonesia began experiencing the disruption from Thursday afternoon towards Maghrib. Users attempting to access AI services for work or to complete tasks began reporting errors gradually.
This disruption was felt extensively by users in Indonesia. Data from Downdetector showed a surge in disruption reports from Southeast Asia and Indonesia specifically. Many users in the country reported being unable to log in, receiving no answers, or encountering error messages when attempting to use ChatGPT, Gemini, or Claude.
This implies that the disruption was not merely a problem in the United States, but a global outage felt simultaneously in Indonesia. Users in Jakarta, Surabaya, Medan, and other major cities reported the same issues, namely services being unresponsive or running extremely slowly.
Based on data recorded on Downdetector, the pattern of disruption was very clear and simultaneous across various continents:
OpenAI / ChatGPT: Reports of disruptions surged very high suddenly around 16:30-17:00 WIB. OpenAI’s official status board noted an “increased error rate across ChatGPT and Codex services.” Reports began to decrease around 19:00 WIB as recovery progressed.
Google / Gemini: The disruption was recorded as widespread and lasted longer, with issues still visible until 21:00 WIB.
Anthropic / Claude: Only the top Opus model was most severely affected from 16:45 WIB, while other models returned to normal faster.
X / Grok: Reports of widespread disruption were recorded from 17:00 WIB and began to subside around 18:30 WIB.
The surge in disruption reports rose and fell in a very similar pattern across various time zones. This pattern strengthens the suspicion that the disruption likely stemmed from the same global cause, rather than being a mere coincidence.
This unexpected event triggered a wave of jokes on social media. Many people realised how much humans depend on AI, to the point that when the services went down, it felt as though the world had stopped spinning.
“Finally, I can feel the sunlight!” wrote one social media user.
“In that very brief moment, millions of people were forced to use their own brains again,” wrote technology observer Paris Marx with a sarcastic tone.
The most shared question was: “If we rent our brains from machines, what happens when the world’s brain rental service goes dark all at once?”
Services began to recover one by one starting from 18:30 WIB, and most had returned to normal by approximately 22:00 WIB.