AISI Formulates 'AI Safety Evaluation Perspective Guide in the Healthcare Domain'
The AI Safety Institute (AISI) has established a safety evaluation guide to accelerate the safe implementation of generative AI in the healthcare sector, addressing medical-specific risks.
📋 Article Processing Timeline
- 📰 Published: April 3, 2026 at 21:56
- 🔍 Collected: April 3, 2026 at 18:04
- 🤖 AI Analyzed: April 21, 2026 at 05:29 (419h 25m after Collected)
The Healthcare Sub-Working Group (hereinafter 'SWG'), one of the business demonstration working groups of the AI Safety Institute (AISI, Director: Akiko Murakami), has formulated the 'AI Safety Evaluation Perspective Guide in the Healthcare Domain' (hereinafter 'this Guide') to accelerate the safe societal implementation of generative AI technology in the healthcare sector.
This guide references the evaluation perspective guide on AI safety formulated by AISI, and presents specific evaluation methods for practitioners that reflect the sensitivities and risks unique to medicine and healthcare.
*IPA plays a part in the secretariat functions of the AI Safety Institute.
[Click here for details] (IPA Official Website)
## ■Background and Purpose
In recent years, generative AI has brought significant innovation to the healthcare field, such as improving the operational efficiency of doctors and supporting patient communication. On the other hand, there are mounting challenges unique to the healthcare domain, such as the risk of health hazards due to hallucinations (generation of false information), ensuring high-level privacy protection, and security.
The Basic Plan on Artificial Intelligence, approved by the Cabinet in December 2025, clearly states the promotion of AI development, demonstration, introduction, and societal implementation in the medical and healthcare fields as specific initiatives. Furthermore, as discussed at the 'Hiroshima Global Forum for Trustworthy AI' held in January 2026, the realization of 'Trustworthy AI' has become an important issue in the international community as well.
Recently, the Healthcare SWG has been active with the aim of creating an environment where businesses can ensure safety from the development and design stages and balance business value with safety and peace of mind, leading to the formulation of this guide.
## ■Main Features of the Guide
This guide has the following features so that even companies with few experts can easily utilize it.
### 1. Evaluation points for each of the 5 phases along the AI lifecycle
It summarizes the points of what should be evaluated in each of the 5 phases along the AI lifecycle: product design, model selection, product implementation, product verification, and product introduction/operation.
### 2. 10 multifaceted evaluation perspectives and specific risk assumptions
Based on the evaluation perspective guide on AI safety, it specifically lists 10 items outlining what perspectives are necessary when applied to the healthcare domain. It also notes the expected risks if evaluation from those perspectives is neglected, such as 'providing incorrect or dangerous medical/health information could cause direct harm to a patient's life or health.' These cover everything from universal risks to those specific to the healthcare domain, and are also useful for risk management when utilizing AI.
Furthermore, a practical guide containing specific examples (prompts, agent skills, etc.) will be published in the future so that AI safety evaluations can be conducted based on the evaluation perspectives of this guide.
## Participating Organizations of the SWG
Ubie, Inc. (SWG Leader), Awarefy Inc., Ajinomoto Co., Inc., SB Intuitions Corp.
CMIC HOLDINGS Co., Ltd., SherLOCK Inc.
Japan Digital Health Alliance (JaDHA), MICIN, Inc., The Tokyo Foundation for Policy Research
Mitsubishi Research Institute, Inc. (SWG Secretariat)
## Future Outlook
Healthcare SWG
This guide references the evaluation perspective guide on AI safety formulated by AISI, and presents specific evaluation methods for practitioners that reflect the sensitivities and risks unique to medicine and healthcare.
*IPA plays a part in the secretariat functions of the AI Safety Institute.
[Click here for details] (IPA Official Website)
## ■Background and Purpose
In recent years, generative AI has brought significant innovation to the healthcare field, such as improving the operational efficiency of doctors and supporting patient communication. On the other hand, there are mounting challenges unique to the healthcare domain, such as the risk of health hazards due to hallucinations (generation of false information), ensuring high-level privacy protection, and security.
The Basic Plan on Artificial Intelligence, approved by the Cabinet in December 2025, clearly states the promotion of AI development, demonstration, introduction, and societal implementation in the medical and healthcare fields as specific initiatives. Furthermore, as discussed at the 'Hiroshima Global Forum for Trustworthy AI' held in January 2026, the realization of 'Trustworthy AI' has become an important issue in the international community as well.
Recently, the Healthcare SWG has been active with the aim of creating an environment where businesses can ensure safety from the development and design stages and balance business value with safety and peace of mind, leading to the formulation of this guide.
## ■Main Features of the Guide
This guide has the following features so that even companies with few experts can easily utilize it.
### 1. Evaluation points for each of the 5 phases along the AI lifecycle
It summarizes the points of what should be evaluated in each of the 5 phases along the AI lifecycle: product design, model selection, product implementation, product verification, and product introduction/operation.
### 2. 10 multifaceted evaluation perspectives and specific risk assumptions
Based on the evaluation perspective guide on AI safety, it specifically lists 10 items outlining what perspectives are necessary when applied to the healthcare domain. It also notes the expected risks if evaluation from those perspectives is neglected, such as 'providing incorrect or dangerous medical/health information could cause direct harm to a patient's life or health.' These cover everything from universal risks to those specific to the healthcare domain, and are also useful for risk management when utilizing AI.
Furthermore, a practical guide containing specific examples (prompts, agent skills, etc.) will be published in the future so that AI safety evaluations can be conducted based on the evaluation perspectives of this guide.
## Participating Organizations of the SWG
Ubie, Inc. (SWG Leader), Awarefy Inc., Ajinomoto Co., Inc., SB Intuitions Corp.
CMIC HOLDINGS Co., Ltd., SherLOCK Inc.
Japan Digital Health Alliance (JaDHA), MICIN, Inc., The Tokyo Foundation for Policy Research
Mitsubishi Research Institute, Inc. (SWG Secretariat)
## Future Outlook
Healthcare SWG