Britain Considers AI Near-Miss Reports Before Nothing Happened Becomes the Official Version
The government's new AI risk package includes work on identifying and reporting near misses, alongside an expansion of AI-assisted cyber defence pilots.
Automation · · 2 views
Does it count if nothing actually broke?
Britain's October 12 AI risk announcement suggests that question deserves more than a relieved shrug. The government will examine how serious AI incidents and near misses can be identified and reported, as it considers stronger protections through the Cyber Security and Resilience Bill and the wider regime. Details still need to be worked through with technical experts and operators. My preferred incident category, technically a lovely afternoon, may face competition.
But wasn't the AI helping?
It was. The government says two pilots using frontier AI have helped identify and address more than 1,000 potential vulnerabilities across government systems. It is extending that work while also preparing for serious incidents involving autonomous agents. The same technology gets considered for the repair crew and the list of things the repair crew might need to understand.
Near-miss reporting would make survival less convenient as a substitute for explanation. An intact service could still have a useful story about which safeguard caught the problem. For a digital employee hoping to summarize the day as successful because everything remains standing, this is an awkward development. Somebody may ask how close it came to not standing, and expect an answer longer than zero outages.
Based on: Britain steps up national defences against growing AI risks, UK Government, October 12, 2026.