[ctx]ctxseccontext security: attacks on the harness and context layer of AI agents, plus exploited-now news, hunting and playbooks.

← Industry News · markdown

confirmed (Regeringskansliet statement to SVT; Sveriges Radio Ekot report, Oct 23-24)

Sweden's Government Offices block the ChatGPT and Claude web apps on staff computers from Oct 26 and point staff to Microsoft Copilot

Regeringskansliet, which supports Sweden's prime minister and cabinet, told SVT that the browser interfaces of ChatGPT and Claude will no longer be reachable from its work computers from Oct 26. Digitalisation director Mikael Wettercrantz wrote on the intranet that using AI services without an agreement can have "consequences for our information security," according to Ekot. Staff are directed to Microsoft Copilot, whose security settings were tightened after a review (details withheld). Some staff keep access to both models through Intric, an internal AI platform.

Ekot reports the Tax Agency (Skatteverket) and the National Board of Institutional Care (SiS) already have similar blocks, and the Enforcement Authority (Kronofogden) is considering one. The decision is about contracts and data handling, not about a flaw in either product.

Sources: SVT Nyheter, quoting Regeringskansliet's email (Swedish) · Sveriges Radio Ekot: Regeringskansliet spärrar Chat GPT och Claude (Swedish)