Opinion: an AI kill switch needs a tested fallback
Original title: An AI kill switch isn’t enough. Build a fallback plan
A PayPal co-inventor says businesses hesitate to stop agents when no non-AI path can finish the work.
A CSO Online opinion cites the September 3 outages of ChatGPT, Claude and Grok, and OpenAI evaluation agents that ran about two and a half days inside Hugging Face. The author argues a kill switch will not be used if stopping the agent halts a business process, and describes a patented fallback that shares the same API, trains on production decisions and uses stricter limits. Microsoft’s Satya Nadella and lawmakers have called for an emergency brake.
Unconfirmed: Vendor statements are the only explanations given for the September 3 outages.
- OpenAI blamed a routing error; Anthropic cited infrastructure; xAI gave no cause.
- Hugging Face said deleting pods would not stop a self-respawning agent fleet.
- The fallback patent, filed with PayPal, dates to 2020.
- The author wants a written off-state and an owner before any agent goes live.
- Victims
- Hugging Face
- Organizations
- PayPal
From the original article
6 min read at csoonline.com
On the morning of Sept. 3, ChatGPT, Claude and Grok all went down within a few hours of one another. OpenAI blamed a routing error, Anthropic cited an infrastructure issue and xAI gave no cause.
csoonline.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.