A Closer Look at Examples of AI Usage Guidelines for State Governments: Pennsylvania and New Jersey
- Angela Novelli
- Jun 20
- 5 min read
The use of AI in business and organizational operations has become increasingly commonplace, with more than 78% of organizations using AI in at least one function. This number will only continue to grow as awareness of the potential in productivity due to AI increases across industries, highlighting the importance of ensuring the existence of AI usage guidelines.
What is the Purpose of AI Usage Guidelines?
AI usage guidelines, also referred to as an AI policy, ensure that members of a company are clear on what the rules are when it comes to utilizing AI tools. It also ensures proper compliance, not only with company guidelines but also legal ones, and outlines how AI should and should not be used.
Having an AI policy helps organizations stay true to the safe and responsible use of AI. Many employees use AI tools without the oversight of their company, also known as shadow AI, which can increase the risk for data privacy issues and a loss of trust with clients and customers. In addition, according to McKinsey’s Global Survey on the State of AI, 30% of organizations only review about 20% or less of generative AI outputs before use. This means that many results from generative AI tools are being made public without review, which could cause risk of inaccuracies and falsehoods.
Without AI usage guidelines, companies are at risk of potential bias or inaccuracies, unintentional data breaches, and legal issues due to a lack of compliance. If your company is using AI or thinking of doing so, take a look at other organizations’ AI policies to help create one of your own. From startups to major corporations and state governments, it is important to explore rules for safe and responsible AI usage. Let’s take a closer look at examples of AI usage guidelines for state governments. In this article, we will focus on the policies for Pennsylvania and New Jersey.
Pennsylvania’s AI Usage Guidelines
The Commonwealth of Pennsylvania has outlined several factors in their AI policy for the use and integration of AI in any government office or agency. Here we will highlight a few key parts of this policy and discuss their importance.
Readiness Assessment: This policy lists that any agencies looking to utilize AI will conduct a readiness assessment in order to ensure a general understanding of AI. In any AI implementation project, this is a very important factor, as those that are working with AI should always have knowledge of what it is and how it works.
Governance: PA’s AI policy highlights governance requirements, ensuring that “overall impact to business and technology operations are not negatively impacted by the integration of AI solutions.” This means that there will be consistent oversight of AI technology, examination of its impact, and actions taken to mitigate any possible risks that could arise. This is key when it comes to using AI in any organization, as always having humans in the loop is a crucial aspect of responsible AI use. It is important to remember that AI should not be allowed to operate completely on its own without intermittent oversight and updates.
Data Management: This ensures that AI is being trained on proper data that is scaled, formatted, and representative of the problem it is used to solve. It is prohibited to use data that is biased, inaccurate, or incomplete, and it is required to put the data through a vetting process. When it comes to using AI tools and programs, the goal is to receive the most accurate and relevant results. This all depends on the data it is trained on, so having this guideline in place is paramount.
Disclosure: Transparency is critical when it comes to AI use, especially in the public sector. Not only is transparency the ethical option, it helps keep trust with any customers, residents, and other individuals that might be viewing something created using AI and not know it. Agencies must disclose that any aspect of publicly available content or information was generated using AI. This disclosure must be prominently displayed and include the system and version of generative AI that was used.
Prohibited Uses: This set of guidelines includes prohibited uses of AI, which is very useful and provides clear directions for public sector users. Some things that are not allowed include the generation of non-text-based outputs including images and video, the use of sensitive and personal information in any public generative AI tool, and using AI for final decision-making, such as a decision related to employment. These are great to include in this policy so that there is no question as to what is prohibited. This list of restrictions is highly important when it comes to state government, especially banning the use of sensitive information in AI tools.
New Jersey’s AI Usage Guidelines
In comparison to Pennsylvania’s AI policy, New Jersey has many of the same ideas in mind. Let’s take a look at some of the features of this set of guidelines.
Fact-Check: This ensures that AI users verify any AI-generated content and fix inaccuracies due to outdated information. It also enforces the need to review for any biased or harmful content or information that can affect vulnerable populations, highlighting the obligation to use AI tools responsibly. This should be a commonplace factor in any organization’s AI policy, as technology will always have the potential to produce inaccurate or biased results. This can be mitigated by consistent oversight and management of the data that AI is trained on.
Disclose: Similarly to PA’s policy, content from NJ government agencies that is in any way created by AI has to be labelled as such. As stated previously, transparency when it comes to the use of AI is important, especially when it comes to public trust.
Sensitive Information: This is typically included in any AI policy, but the public sector has access to large amounts of sensitive information from thousands of individuals. It is highly enforced as a reminder to never include sensitive information when prompting AI tools, including personally identifiable information, content about sensitive topics or meetings, and any other information that you would not share publicly.
This set of guidelines does not go into detail about things that are prohibited, as we saw with Pennsylvania’s. There is also much less included in this policy, excluding details on governance and readiness. The more guidelines included in AI policies, the better. It is much more beneficial to have more aspects included than risking having too little, now that AI is continuing to evolve and its use becoming more widespread across many states in the U.S.
Looking to generate an AI policy for your organization? At Sedna, we have built a team of industry leading subject matter experts in Artificial Intelligence use cases and implementations in organizations that can help you ensure that you have everything covered in your policy so that you can get the most out of your AI implementation. Contact us today to learn more.
“The single most important ingredient in the recipe for success is transparency because transparency builds trust.”
– Denise Morrison, Former CEO, Campbell Soup Co.
Sources:
Comentários