Politics

Copy article

AI in Hospitals Is 'Too Fast', Warns WHO

ended 20. November 2025

Health Policy Watch reports robots and clever computer programmes are transforming medicine, helping doctors diagnose illnesses faster and tracking diseases better However, the world's top health experts are shouting a warning: this technology is moving too quickly, and the safety rules cannot keep up.

The World Health Organisation's European Region (WHO/EURO) released a major report today that said Artificial Intelligence (AI) in hospitals is rapidly advancing past the ethical and legal safeguards meant to protect people. This is creating a “dangerous gap in patient safety.”.

One of the biggest problems the WHO found is that the people who make the rules about AI are not listening to patients. When governments were developing policies for AI in health, the report found that most nations spoke to the AI developers and healthcare providers.

  • 0nly 42 per cent of countries in the region included patient associations in the conversation
  • Just 22 per cent of countries bothered to consult the general public. 

The WHO warned that this limited engagement could mean the new AI tools being developed might not actually meet the real-world needs of patients.

This alarming news comes as the European Commission (EC) actually proposed loosening some of the rules on AI in the European Union.

Critics of the EC's plan warn that easing these regulations, especially those relating to data protection, could allow large companies to use highly sensitive personal information, potentially including people's DNA, to train their AI systems under a broad legal definition. 

This has led one civil liberties organisation to call the proposed changes the "greatest setback for digital fundamental rights" in the EU's history.

We want your views:

  • The report found that only 22 per cent of countries consulted the general public on AI rules. Why did governments forget to ask the patients how they want AI to treat them?
  • If governments ignore the public, will AI programmes be developed that only help richer people or miss the needs of ordinary families?
  • Is it right that only 42 per cent of countries spoke to patient groups? Does this mean the rules are being made for the patients, but not with the patients?

 

2 responses from the Newspage community

Copy all

Star Quote
Copy

Colette Mason
AI Ethics Consultant at Clever Clogs AI
This foolish speed will cost a fortune. In the world of technology, there is a golden rule about fixing errors, often known as the 1:10:100 rule.This principle is widely accepted across the tech industry. It means this:

It costs £1 to fix a problem when the system is still an idea on paper. It costs £10 to fix it when it has been coded and is being developed. It costs a disastrous £100 to fix it after the programme has been deployed and is being used by real users.

The WHO report just showed us exactly how expensive this gamble will be.

Usability is the fancy word for making sure a computer programme is easy, valuable and safe to use. If you do not ask the people who will actually use the system, the patients as well as the doctors, you're guaranteed not find the problems until it is too late.

When only 22 per cent of the public is consulted, you end up with systems that are not built for everyone. This is how you get algorithmic bias and disappointing functionality and results.
Copy

Simon Jones
Chief Operating Officer at VP MED Ventures
This could create a dangerous patient safety gap. When only developers and providers are consulted, the resulting AI tools risk being technologically advanced but clinically misaligned, failing to address the real-world needs and vulnerabilities of the patients they are meant to serve. This oversight could lead to algorithmic bias, where systems are not tested on diverse populations, potentially worsening health disparities for marginalised groups.

Furthermore, this approach is financially reckless. Excluding users during the design phase ensures that critical flaws will only be discovered after deployment, when fixes are exponentially more costly and the potential for human harm is greatest.

Ultimately, creating health technology for patients but not with them is ethically indefensible and operationally unsustainable. It prioritises speed and corporate interests over fundamental rights and safety, risking a catastrophic loss of public trust in both healthcare institutions and the te