Elevated Errors in Claude.ai

Claude.ai had a wobbly morning, with errors across its site, developer platform, and coding tool before a fix was pushed and the team said they’re monitoring. But the real action was the comments, where users went full town hall. One voice called out “98.92% uptime” as “horrendous”, blasting the idea of “one nine” (around 99%) as nowhere near the “five nines” (99.999%) folks dream about. Others countered that a sudden stampede of new users—reportedly a “flock from OpenAI,” according to a Bun engineer—would rattle any system.

Meanwhile, everyday users told their tales. One person’s long incognito chat with Claude vanished mid‑conversation, leading to a DIY séance: could reposting the transcript resurrect the chat, like they’d done with Google’s Gemini? Then came the jokes. “Who fixes the AI when the AI is down?” one quipped, poking at the myth that you can run a modern service without writing much code. The spiciest suggestion: keep an emergency backup Claude to fix production Claude— or even, gasp, have Claude call in OpenAI or Gemini when it breaks. Big yikes, big laughs.

So yes, the status page moved from “investigating” to “fix in place,” but the court of public opinion is still in session: reliability vs. scale, panic vs. patience, and a whole lot of memes about a robot calling another robot for help.