The rapid increase in the use and ability of Generative AI offers a sea of possibilities for Washington ports. From drafting tenant communications to summarizing complex shipping data, tools like ChatGPT and Microsoft Copilot are intended to make our daily work more efficient. However, as public agencies, ports operate under a microscope of transparency that private companies do not. The efficiency of AI cannot come at the cost of public trust or legal compliance.

For port officials, records officers, and legal teams, the integration of AI brings a pressing question: How does this new technology fit into the established framework of the Washington Public Records Act (PRA)? The answer has uncertainty. While most generative AI outputs are likely public records, there may also be more nuanced intersections of generative AI and the PRA, including authorized tools, inputs, retention, and sensitive information, where there is greater uncertainty as to what is a public record.

Is AI Content a Public Record?

The short answer is: Yes, it certainly can be.

Under RCW 42.56.010, a public record is any “writing” containing information relating to the conduct of government or the performance of any governmental or proprietary function that is “prepared, owned, used, or retained” by any state or local agency. This broad definition includes digital information captured by agency systems while employees go about their duties.

Courts in Washington have historically taken a wide view of what constitutes a record. For instance, in Belenski v. Jefferson County (2015), the court found that automatically generated “internet access logs”—which simply tracked when an agency computer accessed the web—were public records. If a passive log is a public record, a deliberate prompt typed by a public employee to generate government content might be as well. In fact, the State of Washington’s guidelines for local governments specifically direct agencies that the definition of public records under the PRA encompasses content created with generative AI systems.

Therefore, if a port employee uses a generative AI tool to draft a press release about a new terminal opening, create an email to a commissioner, or summarize environmental impact data, the resulting text could be a public record if it relates to the conduct of government and is being “used” by the agency.

When people think about AI, they often focus on the output—the essay, the image, or the code it produces. However, Washington’s guidelines indicate that from a PRA perspective, the output is only half the story.  The Washington State Archivist suggests that generative AI inputs—the specific prompts, questions, and data fed into the AI—could also be public records.

The Hidden Trap: Retention

In the context of the PRA, content and function, not form, dictates retention. So while ports can likely apply existing retention schedules to AI outputs and inputs, storage must also be addressed.

It is unlikely that prompts, on their own, need to be retained. Prompts may be equivalent to internet browser history, which is considered transitory and does not have a record retention schedule under the Washington Secretary of State’s Local Government Common Records Retention Schedule (CORE). If generative AI prompts are used as a sounding board, such logs “can be retained under the transitory Brainstorming and Collaborating DAN,” according to a Record Management Advice issued in 2024 by the Washington State Archivist.

However, when prompts produce an output that is used by a port employee, the content of the output will dictate the retention schedule that applies. Like public records in any format, there is a spectrum of retention requirements to consider.

To determine the content/function of generative AI records, the State Archivist recommends a person ask:

1. What is the interaction about? (content)
2. Why did it take place and for what purpose? (function)

Refer to the Local Government Common Records Retention Schedule (CORE) to determine the appropriate lifecycle for the content you create.

Implications for Ports

The intersection of AI and the PRA creates several specific challenges ports must manage.

1. Storage

Since most standard AI tools don’t currently integrate with municipal archiving software, employees may need a manual workflow to ensure retention. If an employee has a conversation with ChatGPT in a web browser and then closes the tab, that record might be lost forever unless specific steps are taken to save it. This could create a risk of unlawful destruction of public records. Using a designated account will save chats and may serve to remind employees of their obligation to retain such records. To avoid incomplete recorded responses, port policies might contemplate not only the use of authorized tools but also whether employees must log in to ensure chat history is saved.

2. Searchability and E-Discovery

When a records request arrives, the Records Officer must be able to search for responsive documents. If AI interactions are scattered across individual user accounts on third-party platforms, the Records Officer may not be able to easily search them. This may create a potential compliance gap where responsive records exist but are not located or produced.

3. Third-Party Data Privacy

Ports manage proprietary data, tenant leases, security protocols, and environmental studies. If a port employee inputs sensitive tenant financial data or security protocols into a public AI model to get a summary, that information may be ingested by the AI company. While this is primarily a security and privacy issue, it complicates records management if that input contains exempt information that is now residing on a third-party server.

Navigating Forward

As technology eclipses legal frameworks, ports are in uncharted waters navigating how new technology is used and how to continue to uphold the trust of the communities they serve.

While it is not clear how AI-generated content will be treated under the PRA, ports might consider enacting generative AI policies that address both inputs and outputs, which tools are approved for use, how sensitive and confidential information can be used, and how information is stored and retained. And certainly, ports can still continue to analyze the recorded responses for possible exemptions.

This article summarizes aspects of the law and opinions that are solely those of the authors. This article does not constitute legal advice. For legal advice regarding your situation, you should contact an attorney.