Arctera has found that UK organisations are outpacing their global peers in embedding AI governance into daily operations, yet many cannot demonstrate how an AI-assisted decision or communication came about when it is challenged.

The findings appear in Arctera’s latest study, State of AI Governance 2026: The UK Perspective – Where Operational Governance Meets the Reconstruction Challenge. UK respondents reported higher adoption than the global average across every major control measured.

Some 89% insist on human sign-off for high-risk AI communications, against 79% worldwide, while 85% train staff on compliant AI usage compared with 77% globally. A formal AI usage policy exists at 83% of UK firms versus 80% globally, 77% track or log AI prompts and outputs against 65% elsewhere, and 70% carry out routine audits of AI tools and outputs, compared with 63% globally.

Arctera cautions that deploying controls addresses only half the problem. Firms must also show regulators, auditors, legal teams and internal investigators concrete proof that those safeguards functioned as intended once an audit, probe or dispute arises.

That task is complicated by fragmentation. The records generated by AI controls frequently sit across separate platforms, departments and channels, and 40% of UK respondents named linking communications across formats and channels as a significant obstacle.

Exposure rises sharply where staff rely on consumer messaging apps or personal calls: 73% of UK firms describe their monitoring of these channels as partial or absent, while the equivalent figures stand at 59% for personal phone calls and 27% for text messages.

Source link