Claude experienced two large-scale outages within two days, while Anthropic reported facing "unprecedented demand" BlockBeats reported that on March 3rd, Anthropic's AI assistant Claude experienced a large-scale service interruption for two consecutive days on March 2nd and 3rd. On the morning of March 2nd at 6:49 AM Eastern Time, Anthropic announced that it was investigating the issue of "increased error rates" in claude.ai, mobile applications, Claude Opus 4.6, Claude Console, and Claude Code. Downdetector data showed that over 2000 users reported faults during peak hours. After the service was restored, Anthropic stated that the team is "working hard to match Claude's huge demand in recent days". On March 3rd at 03:15 UTC, the official status page once again reported an increase in error rates for claude.ai, cowork, platform, and Claude Code. As of 04:43 UTC, they are still under investigation. In the two crashes, both Claude API (app. (anthropoic. com)) and Claude for Government were not affected, while consumer services were the most impacted. Anthropic attributed the malfunction to "unprecedented demand," and according to its disclosure, the number of free users has increased by over 60% since January this year, while the number of paid subscribers has more than doubled since October last year. Bloomberg reported that the surge in demand was triggered by user backlash after OpenAI signed an AI cooperation contract with the US Department of Defense - ChatGPT's uninstallation volume in the US surged 295% month on month, and one star ratings surged 775%, while Claude's US downloads increased by 37% -88% month on month, surpassing ChatGPT's total daily downloads at one point. Users can view the real-time status of each service component through status. (claude. com).