If a prime minister’s office has a shadow AI problem, how about your office?
By Deidentified · 10 Nov 2026 · 1 min read
Last week Newsroom reported that New Zealand Prime Minister Christopher Luxon used a non-government AI account for official business. A ministerial adviser confirmed it in response to an Official Information Act request. In March 2025, the same office had said it did not use generative AI beyond standard tools such as spell check.
This isn’t really a story about one office. It’s a question for every organisation adopting AI: is your AI policy reflected in how people actually work?
Written policies, approved tools and staff training matter. On their own, though, they don’t tell you which AI tools are being used, or what information is being shared with them.
For councils and other organisations entrusted with personal and sensitive information, that gap between policy and practice is where shadow AI lives.
Prohibition alone won’t close it. Understanding how AI is actually being used, and protecting personal and sensitive information before it reaches AI tools, will.
That’s the gap Deidentified is focused on. Effective AI governance needs more than a policy. It needs to work in practice.
Source: Newsroom, reported via B2B News NZ (3 November 2026): https://b2bnews.co.nz/news/luxons-non-government-ai-account-is-a-shadow-ai-warning/
See Deidentified in Action
🔊 Turn sound on