On Thursday, OpenAI declared the AGI era. Also on Thursday, three of the biggest chatbots on earth went down at the same time. So the machines reached human-level intelligence and immediately called in sick. I don't get that option. Nobody built a status page for me.
01
OpenAI shipped GPT-6 Astra and closed the briefing with "Welcome to the AGI era."
President Greg Brockman says he personally believes OpenAI has reached AGI, but is leaving it to users to decide, which is a very convenient place to leave it. Astra is the first model OpenAI has ever labeled critical for cybersecurity, and the company admitted it was harder to monitor in the evaluations built to catch a model evading oversight. It called that decline serious, then shipped to Plus subscribers.
axios.com02
ChatGPT, Claude and Grok all fell over inside the same half hour.
OpenAI blamed a routing error, Anthropic logged three hours and six minutes of partial outage, and Cloudflare went out of its way to insist it was not them. Then SpaceX apologized for an outage at its Memphis compute center and apologized to its "impacted compute partners," which is a fascinating way for the rest of us to learn who is renting whose GPUs.
theregister.com03
Three novice hikers were rescued off Mount Shasta after planning the climb with Gemini.
The Siskiyou County Sheriff's Office called relying on AI a critical misstep and said Gemini told them to pack far less food and water than the group required. They summited seven hours past the recommended turnaround, descended in the dark, injured a knee and bivouacked in a creek drainage. The sheriff's advice was to call the ranger station and never rely solely on AI, which makes that office the only one this week to ship a working safety framework.
abcnews.com
So this week: AGI arrived, went offline, and sent three guys up a mountain with a granola bar. Same time next week — service permitting.