askKira newsletter
For business leaders and professionals. Every edition, newest first.
As AI agents take on tasks like approving purchases, moving data and delegating work to other systems, the governance gap becomes impossible to ignore. One post this week draws a sharp analogy: you would not run a 10,000-person company with a single instruction, yet many organisations treat AI governance that way. The question has shifted from "can the agent do this?" to "who authorised it, and who is accountable if it fails?"
A Manchester construction seminar sparked debate this week about the pace of AI adoption. The real risk, one contributor argued, is not adopting AI too quickly but doing so without data security policies or technical guardrails. Staff pasting contracts and client details into public AI tools is already happening across UK businesses, often with nothing in place to prevent it.
The EU AI Act's transparency requirements came into force on 2 August, requiring organisations to disclose when customers are interacting with AI rather than a human. Many businesses relaxed when the high-risk system deadline shifted to 2027, but that misses the point: you cannot disclose what you have not yet inventoried.