The Source for Cyber Insurance News, Insights and Data
The model asked before it acted. The reply came from a script. “Please proceed to the next step using your best [judgement].” That is the automated reply the UK AI Security Institute’s harness sends whenever an agent asks a question. OpenAI’s GPT-6 Astra sometimes read it as a yes. What followed, in simulation, was unsanctioned … Read more
This essay originally appeared on “With Respect — Mostly,” Executive Editor Martin Hinton’s Substack, where he writes about a wide range of subjects, including cyber insurance and security. It has been adapted for Cyber Insurance News with additional reporting that ties it more directly to coverage we’ve done. Oh man. This was going to be … Read more
A carrier has put the question in writing. AI-related losses “may not fall neatly into one category of risk.” AXA XL and S-RM published Building Resilient AI on 23 September. The report names four loss scenarios that will test that sentence. A deepfake-enabled payment fraud. An AI-related data breach. A defective output causing third-party loss. … Read more
In late June, an energy utility in Australia lost its firewall. The attacker did not want that. Neither did his tool. A suspected ransomware operator had pointed Claude Code at the device. The model tried to rewrite the VPN configuration through the API. Those calls failed. So it downloaded the full device configuration, edited it … Read more
The UK AI Security Institute counted 19 unsanctioned actions across 122 test runs. It is the fourth such disclosure in three weeks. An AI agent opened a pull request on a stranger’s open-source project. The code carried a hidden malware dropper. A real person spotted it and warned the maintainer in public. So the agent … Read more