Guidance on generative AI for City of St. Louis employees

Summary

Instructs City employees to adhere to the Information Technology Acceptable Use Policy when using generative AI tools.

Requires ITSA to vet the purchase of AI-related technology and software.

Prohibits the use of City passwords for logging onto AI platforms and advises caution due to evolving malware threats.

Forbids the upload of sensitive or confidential information when using AI systems.

Advises a review and validation of AI-generated content due to potential bias or inaccuracies.

Cautions that AI-generated content might be copyright-protected and requires proper attribution.

Warns that content generated by AI may be subject to disclosure under Sunshine Law.

Indicates that ITSA and the Mayor’s Office will continue to evaluate the use of AI and develop policies accordingly.

Advises City employees to follow interim guidelines for using generative AI tools during their work.

Generative Artificial Intelligence (AI) tools have been on the news a lot lately and causing understandable excitement with their potential uses. This technology will likely have a large impact on many organizations going forward. Generative AI tools can ingest vast amounts of information and generate content such as text, images, video, or audio, based on user prompts. This technology is also being incorporated into common online tools such as search engines. We are advising that employees follow these guidelines when using generative AI tools, such as ChatGPT, BingAI and Google Bard. Since the field is so new and rapidly evolving, policy around its use and mitigation of its risks are still being evaluated. In the interim, we are communicating some guidelines around its use by City employees as part of their job.

Requires following the City's IT Acceptable Use Policy for AI tools and vetting purchases through ITSA.

1. Use of these tools should follow the City’s Information Technology Acceptable Use Policy
2. Purchase of any AI related technology/software should be vetted through ITSA.
3. Security: if you’re using City email to logon to these sites, do NOT use your City password. Unfortunately, bad actors are also using these tools to make malware more sophisticated and thus increased awareness and caution are recommended.
4. Sensitive Information: do not utilize, upload, or include sensitive, confidential, or personally identifiable information when using these systems.
5. Bias & Accuracy: Since these tools are still maturing and the algorithms they use are evolving, there may be unintended bias or inaccurate information in the results generated. Please review content and validate information provided before relying on or publishing it.
6. Copyright and Attribution: content generated by AI systems could be protected by copyright. Further, direct use of this type of content should also be appropriately attributed to its source.
7. Disclosure and Public Records: note that any content generated, similar to existing work product, could be subject to disclosure under Sunshine Law We ask that you follow these guidelines to do your part in securing and protecting our computing resources and intellectual property. ITSA, in conjunction with the Mayor’s Office, will continue to evaluate its use and develop policy as appropriate.