Unmanaged AI Raises Patient Safety Risks as Singapore Hospitals Struggle to Keep Pace

Singapore's healthcare sector is facing a growing but largely invisible problem: doctors and nurses are using artificial intelligence tools faster than hospitals can govern them. According to a new report from Nutanix, 79 per cent of healthcare organisations in Singapore are dealing with unmanaged AI adoption, and 83 per cent recognise the resulting business risk. The findings point to a tension that will only intensify as clinical demand for AI tools outpaces institutional controls.

Staff using public AI models for clinical tasks may expose patient information, receive inaccurate outputs, and leave hospitals without an audit trail — risks that are especially acute in Singapore's tightly regulated healthcare environment. Rathanesh Ramasundram, Director at Frost & Sullivan, described the threat as "serious but predictable," driven by clinical pressure and slow access to approved tools. Jay Tuseth, Vice President and General Manager for APJ at Nutanix, said the solution is counterintuitive: hospitals must make approved systems even easier to use than the ungoverned alternatives.

"The answer for the organisation is actually to create the governed model that is easier to use and deploy than the ungoverned model," he said. The challenge is compounded by the unique infrastructure demands of healthcare. A typical intensive-care bed connects to 15 to 20 devices producing continuous data, requiring some processing to remain on-site. Tuseth argued that time-sensitive clinical applications should run on-premise or at the edge, while model training and population-health analytics can use cloud infrastructure — a hybrid model that many Singapore hospitals have yet to fully implement.

Ramasundram advised providers to avoid trying to modernise infrastructure, governance, and workflows simultaneously. Instead, small governed pilots can build clinical evidence before wider investment. The three areas remain linked: frontline workflows deliver value, but only when supported by modern infrastructure and controls covering security, cost, and data sovereignty. The immediate test for Singapore's healthcare system is whether hospitals can give clinicians faster approved tools without weakening accountability or care quality.

Why it matters for Singapore: As Singapore pushes toward its Smart Nation vision and Healthier SG goals, the unmanaged AI problem represents a quiet vulnerability that cuts across care quality, data privacy, and regulatory compliance. The Nutanix data is a reminder that governance infrastructure must scale alongside clinical AI adoption — and that the path to safe AI in healthcare runs through deliberate, governed pilots rather than uncontrolled frontline experimentation.