Current through Register Vol. 54, No. 49, December 7, 2024
Section 7.993 - Core values governing the use of generative AI technology in this Commonwealth(a) An agency shall weigh the design, development, procurement and deployment of generative AI technology based on the following core values: (1)Accuracy-When using generative AI, an agency shall confirm that the generative AI produces accurate and verifiable information. The OA shall develop policies on how best to audit and verify generative AI outputs to ensure clear and accurate information.(2)Adaptability-The fast-evolving nature of generative AI and its potential use cases require an agency to establish and maintain an ability and willingness to recognize and adapt to shifting risks and opportunities. Moving at the speed of business requires the Commonwealth to make investments that promote continued research and diligence; engage with external stakeholders and subject matter experts; and learn from other government partners.(3)Employee empowerment-While using generative AI, an agency shall ensure its employees are both better serving residents of this Commonwealth and improving their own skill set for a changing economy and professional development. The Commonwealth and the OA shall create training opportunities for employees to grow their skills in utilizing, understanding and managing generative AI tools or technology.(4)Equity and fairness-An agency shall use generative AI in a way that does not unlawfully discriminate against or disparately impact individuals or communities based on or due to ancestry, disability, age, sex, national origin, familial status, race, gender, religious creed, color, sexual orientation, gender identity or gender expression. An agency shall use generative AI in a human-centered and equitable manner testing for and protecting against bias so that its use does not favor or disadvantage any demographic group over others.(5)Innovation-The Commonwealth should be a public sector leader in harnessing generative AI to improve the delivery of public services to customers, residents, visitors and industry and to enable an agency to be more effective and efficient in their daily work. An agency shall proactively and iteratively evaluate and present to the board for consideration potential salient use cases of generative AI.(6)Mission aligned-When using generative AI, an agency shall ensure that it is used in service of their core missions to serve customers, residents, visitors and industry, and to support Commonwealth employees.(7)Privacy-The design, development, procurement and deployment of generative AI by an agency must not adversely affect the privacy rights of users. An agency shall ensure that training related to the use of generative AI and the input of data into those tools complies with applicable laws, regulations and policies concerning the privacy rights of users. (8)Proportionality-The OA shall develop policies that encourage or control the use of generative AI in a context-specific manner that is consistent with these core values, but which does not unnecessarily burden end users or an agency.(9)Safety and security-The OA shall lead the development and implementation of standards, procedures and policies to safeguard and secure the data provided to the Commonwealth by customers, residents, visitors and industry against unauthorized uses and intrusions.(10)Transparency-An agency shall ensure transparency and accountability in the design, development, procurement and deployment of generative AI in a manner that respects and strengthens public trust. When using generative AI tools to create content, agency external-facing services or data set inputs or outputs shall disclose the use of generative AI; and what bias testing was done, if any.(b) The OA shall proactively consider the core values listed in subsection (a) in all activities, policies and processes that leverage or intersect with technologies that enable automated decision-making.Adopted by Pennsylvania Bulletin, Vol 53, No. 41. October 14, 2023, effective 9/20/2023