Swiss doctors spend up to two hours a day documenting patient records. Recent data from the FMH 2025 Follow-up Study reveal a worrying trend: excessive bureaucracy and a lack of strategies are slowing down day-to-day medical practice. To escape this pressure, medical professionals are increasingly turning to unauthorized AI tools – so-called «Shadow AI». However, this entails significant security and compliance risks. The following article explains why these tools are still being used and how Voicepoint is addressing this challenge.
Documentation Burden at Record Levels: A Structural Crisis in Day-to-Day Work
The administrative pressure in the Swiss healthcare system is noticeably impacting day-to-day patient care. According to a companion study by gfs.bern commissioned by the FMH, physicians in acute care spend an average of 114 minutes per day working on patient records; in psychiatry, that figure rises to 121 minutes. (1)
Added to this are time-consuming regulatory and insurance-related requirements: In outpatient practice, this takes 54 minutes per day, and in rehabilitation, 67 minutes. The main drivers are inquiries from insurers regarding billing, reports for loss-of-earnings insurance, and formal cost approvals.
This trend is placing a significant strain on healthcare professionals:
- In psychiatry, 51 % of respondents report health-related staff departures within their department over the past 12 months.
- 86 % of psychiatrists cite constant pressure to perform.
- Despite high levels of motivation toward medical work (over 89 % in psychiatry and 91 % in rehabilitation), organizational shortcomings threaten to gradually erode this attitude.
The Phenomenon of Shadow AI: Causes and Risks
Under constant time pressure, professionals seek pragmatic ways to lighten their workload. When institutional software systems are inflexible, this increasingly leads to the use of uncertified consumer tools – known as Shadow AI.
What Is Shadow AI, and Why Is It Used Anyway?
Shadow AI refers to the use of AI tools (e.g., freely available language models or dictation apps) without authorization or review by the organization’s IT and data protection departments. The motivation is not malice, but rather work overload.
Empirical data illustrates the extent of the problem: According to the Doctolib Digital Health Report 2026, 50 % of the doctors surveyed already use private AI tools for research, and 28 % use them directly for treatment documentation. This is offset by clear potential: Certified AI consultation assistants can save up to 70 % of the time spent on documentation. (2)
What Are the Risks of Shadow AI in Healthcare?
Uncontrolled use poses massive risks for healthcare facilities:
- Data protection and compliance violations: When data is entered into public cloud services, sensitive information ends up on external servers.
- Lack of data processing agreements (DPA): Consumer tools lack binding contracts. Providers often reserve the right to use data to train future models.
- Medical misinformation: Generic chatbots are not trained on specialized medical terminology and tend to make statements that sound plausible but are factually incorrect («hallucinations»).
Professional, Certified Alternatives Instead of Shadow AI
To eliminate the risk of Shadow AI, hospitals, clinics, and medical practices need powerful, data-protection-compliant alternatives. As the Swiss market leader in medical dictation solutions, speech recognition, and automated report generation, Voicepoint bridges the gap between efficiency and security and offers:
- Secure AI consultation assistance: Our Voicepoint Xenon® platform generates automatic summaries of doctor-patient conversations.
- 100% Swiss data protection and 3x ISO-certified: Your data remains in Switzerland.
- Seamless integration: Medical reports are imported directly into existing EHR and PIS systems.
- Medical specialization: The language models are precisely tailored to the Swiss healthcare system and medical terminology.
Through certified solutions, healthcare professionals work with us to reclaim valuable work time for direct contact with their patients – without compromising the protection of sensitive personal data.
References
(1) Outpatient Before Inpatient: Process Adjustments Without Strategic Anchoring (fmh.ch).