A new wave of 'Shadow AI' is worrying workplaces (www.techradar.com)

Microsoft research shows a second wave of “shadow AI” is booming in UK workplaces: 71% of employees have used an unsanctioned AI tool at least once and 51% do so weekly. The drivers are practical — 28% of workers report no work-approved AI is provided, so staff turn to consumer tools to draft emails (49%), build reports and presentations (40%) and help with finance tasks (22%) to hit quotas and deadlines. Confidence in workplace AI is rising (57% view it positively), which is accelerating adoption even where policy and tooling don’t exist. That widespread, unsanctioned use is significant because of real security and compliance risks: employees frequently paste company or customer data into external models without sanitizing it, and those prompts could be retained or used to train models and later exposed via clever queries. Only about a third of users are worried about data leakage (32%) or IT security impact (29%), revealing a dangerous awareness gap. The technical implications point to the need for enterprise-grade solutions, governance and tooling — vetted models, prompt-sanitization, data-loss-prevention, logging/audit trails and employee training — so organizations can capture productivity gains without exposing sensitive data.