Adding a Generative Artificial Intelligence (AI) Policy to your Employee Handbook

Generative artificial intelligence (AI) tools like ChatGPT, Bard, Jasper Chat, and others are rapidly gaining popularity. As the world of AI expands and changes, the tools will become more commonplace. As with any technology, generative AI tools come with risks and drawbacks—but they can also be helpful when used properly.

Your employees may already be experimenting with generative AI tools on their own time for their personal interests. Some may also be wondering how they can use it for work. Having an organizational policy in place can equip employees with important knowledge and practices to help harness the potential and minimize the risks of these tools. Establishing an organization-wide policy helps to ensure that innovative AI tools are used responsibly and securely.  

Guidelines

This policy was developed with ChatGPT in mind. If your organization uses different tools, you can modify this policy to accommodate for those tools' unique features and capabilities.

Note: If you are basing your organization’s generative AI policy on an AI program other than ChatGPT, be sure to double check that it has some of the same safety features discussed below.

Additional Article: Is Generative AI Too Good to Be True? Tips for Ministry Leaders

Sample Policy
Using Generative Artificial Intelligence (AI) at Work

Generative artificial intelligence (AI) services can be powerful and valuable tools, but you must be aware of some significant risks. [Name of organization] will permit the responsible use of [ChatGPT and/or other vetted AI tool(s)] as long as employees carefully consider the following issues.

Note: Employees are only permitted to use generative AI tools that have been vetted and approved by church leadership. If you would like to get a tool approved, please contact [administrator name].

Don’t include restricted information in any interaction with generative AI tools.

Since user input is used by default to train the AI models, any information you provide via a query could be made publicly available. This risk will be mitigated by [using ChatGPT exclusively with the “Chat History & Training” setting disabled,] so users’ content won’t be used to train the model. [Presently only ChatGPT offers this limitation option. As other generative AI tools offer this feature, they will be evaluated for employee use.]

The use of generative AI tools is subject to the terms of the Employee Handbook, specifically [name of relevant policies, such as “Confidential Information,” “Student Privacy,” etc.].

Never include sensitive content in queries, such as:

• Personally identifiable information (PII)
• Personal health information (PHI)
• Organization’s intellectual property or trade secrets
• Other confidential attendee, family, student, volunteer, donor, alumni, employee, customer, or third-party information
• Strategic information, including questions that might not appear especially problematic but could inadvertently give away strategic organizational plans (Example: Asking ChatGPT to summarize notes from a board meeting during which your directors discussed financial updates and upcoming changes that are confidential.)

If you’re unsure if your information is safe to enter, seek guidance from your supervisor and human resources administrator.

AI-generated content is not authoritative.
Generative AI tools don’t have the capacity to recall or cite the specific sources from which their responses are generated. Instead of quoting specific phrases, their responses are constructed based on patterns they have learned from a vast array of sources. This means the responses you receive may not be 100% accurate. Therefore, you should confirm the accuracy of the information by cross-referencing the reliable sources you would typically use to complete your assigned tasks or make decisions.

Generative AI tools are meant to enhance human work.
Don’t use these tools as a substitute for the unique skills, insights, and decision-making abilities you bring to your role. Evaluate the quality and relevance of the information provided by AI and use your own expertise and judgment to make informed decisions about what you should include in your work. You should also inform your supervisor if you use generative AI tools as part of the drafting or creation of any content that will be published for consumption outside of [organization name].

Take ethical considerations into account while using generative AI.
All [organization name] employees are expected to adhere to legal and ethical standards while using [ChatGPT]. Avoid engaging in activities that may infringe upon privacy, intellectual property rights, or violate any regulations or laws. Additionally, when creating content of any kind, employees should clearly distinguish between their own input and AI-generated responses and attribute appropriately when sharing information.

Report issues and feedback.
To inform and improve our organization’s practices surrounding generative AI, employees are encouraged to provide feedback on the effectiveness, limitations, or any problematic aspects they encounter while using the system.

These new technologies are rapidly evolving. As we gain more experience with generative AI technology and its capacity to self-verify increases, more guidance will be provided in the future.

When in doubt, err on the side of caution. Always be deliberate and diligent in your efforts to protect [organization name] and the people we serve when using generative AI technology.

Posted July 2023

This is a sample workplace/employee handbook policy only. Your organization is responsible for compliance with all applicable laws. Accordingly, this document should not be used or adopted by your organization without first being reviewed and approved by a licensed attorney in your area. Brotherhood Mutual assumes no liability in the preparation and distribution of this sample document.