Page 7 of 8: Guidelines for using generative AI tools like ChatGPT at Unity Health

Over the past year, generative artificial intelligence (AI) tools like ChatGPT, Microsoft Bing Chat and Google’s Bard have become more popular. To support staff, physicians and learners to responsibly use these tools, we have created provisional guidelines on the use of generative AI at Unity Health. 

Unity Health is the first hospital in Canada with a dedicated program for applied AI. Since 2017, we have developed and implemented over 50 different AI and analytics tools into practice (learn more). AI has the potential to transform and improve health care for provider and patients, but we must also consider the privacy, legal and ethical implications of using these emerging technologies.

Key reminders and actions:

• Do not enter confidential or personal information (including personal health information, employee information and intellectual property) into internet-based tools – including generative AI tools(e.g. ChatGPT, Microsoft Bing Chat and Google’s Bard) – as this is a privacy and IT security risk and would violate the following Unity Health policies:
  • Privacy of Personal Health Information
  • Information Security Policy
  • Secure Handling of Confidential Information in Digital Format Policy
• Do not rely on the accuracy and objectivity of information provided by generative AI tools. Generative AI content must be reviewed and fact-checked before being used, as these tools can produce content that is factually inaccurate and potentially biased.
• Review the new guidelines. Note: the guidelines are less than three pages and should not take more than five minutes to review.

These guidelines were developed with input from IT Security, Privacy and Data Science and Advanced Analytics. They will continue to be updated as new information and requirements surface and as technology rapidly changes.