Andre Sayles becomes Seattle police chief amid open AI-risk

Andre Sayles became Seattle’s interim police chief on July 30, 2026, after Mayor Katie Wilson accepted Chief Shon Barnes’s resignation and named Sayles, then a deputy chief, to lead the department on an interim basis.[1] For readers searching “Andre Sayles Seattle police chief,” that appointment is the immediate answer. The legal-risk answer is narrower: the leadership change does not erase the Seattle Police Department’s open, documented AI-use record.

This is not a personal AI-misconduct allegation against Sayles. The materials cited here do not identify a court sanction, judicial order, or prosecutor finding naming him. The issue is the institutional file he inherits: an OPA management action recommending an AI policy, a city status table still listing that policy work as “In Development” in the latest cited records, two documented officer AI-use misconduct records, and a standing King County prosecutor position refusing AI-assisted police narratives.

Sayles has enough background to orient the personnel news without turning this into a biography. FOX 13 described him as a 20-plus-year law-enforcement veteran, a former Beloit, Wisconsin police chief from 2021 to 2023, and SPD’s Deputy Chief of Administration and Investigations since September 2025.[2] Those facts explain why City Hall could turn to him quickly. They do not answer whether a later report narrative, use-of-force entry, or communications record can be cleanly explained if AI touched it.

The open file Sayles inherits

The central document is OPA Management Action 2024OPA-0333, dated April 4, 2025. In that management action recommendation, OPA stated that SPD Policy 12.040 did “not provide guidance on whether or how any Department employee may use AI” and noted that the city’s Generative AI attribution policy was not referenced in SPD’s Policy Manual. OPA recommended that SPD develop an AI policy.[3]

The status trail matters because it prevents the appointment headline from becoming a legal reset. Seattle’s policy-recommendations page lists the related item, 2025COMP-0007, “Use of Artificial Intelligence,” as “In Development,” with an SPD response dated July 18, 2025.[4] SPD’s response to the management action is separately posted as a July 18, 2025 response document.[5]

There is an important cutoff. The policy-recommendations page reflects documents through March 24, 2026.[4] That means the narrower conclusion is the only safe one: as of Aug. 3, 2026, adoption of an SPD AI-use policy after Sayles’s July 30 appointment is not verifiable from the latest cited city records. It would overstate the file to say no policy could have been adopted after that cutoff. It would also overstate the personnel story to imply the open management action disappeared when the interim chief changed.

Why the non-sustained ChatGPT case is not a clearance

OPA Closed Case 2024OPA-0333, closed Feb. 21, 2025, is the practical reason the policy file is not an abstract governance item. The case involved a sergeant who typed a paragraph into ChatGPT, pasted it into Blue Team, SPD’s use-of-force record system, and encouraged other officers to use ChatGPT for force reports.[6]

OPA did not sustain a city-policy violation because there was no policy to violate.[6] That distinction is the whole point for discovery and risk review. A non-sustained finding based on policy absence is not an affirmative approval of the practice, not a finding that AI-assisted force narratives are reliable, and not a finding that the department had a usable audit trail for later litigation.

For a defense lawyer, prosecutor, records officer, or agency risk manager, the uncomfortable question remains after closure: if a force-report paragraph was drafted, revised, summarized, or polished through a tool outside the ordinary records system, where is that visible in the case file? If the answer depends on voluntary disclosure rather than policy, logging, retention, and attribution rules, the non-sustained closure does not solve the litigation problem.

The second documented AI-use misconduct record is shorter but still relevant. In Closed Case 2024OPA-0178, closed Nov. 8, 2024, SPD fired a student officer for using AI on an academy assignment.[7] That case is not the same as AI-assisted report writing, and it should not be treated as proof of a general frequency. It does show that AI use had already become an internal discipline issue before the April 2025 management action.

The prosecutor’s rule is already clearer than SPD’s policy record

King County prosecutors had already drawn a hard litigation line before the OPA management action. In a Sept. 20, 2024 memo to police chiefs, Chief Deputy Prosecutor Daniel J. Clark wrote that the King County Prosecuting Attorney’s Office “will not accept any police report narratives that have been produced with the assistance of AI. All reports must be produced entirely by the authoring officer.”[8]

The memo was not framed as a technology preference. It identified courtroom consequences: perjury certification concerns, Brady and potential impeachment disclosure issues, the absence of an audit trail proving whether an error came from an officer or an AI system, and a reported example in which an AI-assisted report referenced an officer who was not at the scene.[8]

That prosecutor position changes the weight of the SPD management action. Inside the department, the April 2025 problem was that existing policy did not answer whether and how employees could use AI. In a King County prosecution, the answer for police report narratives was already no: prosecutors said they would not accept them if produced with AI assistance.[8]

Fair and Just Prosecution later characterized King County as the first prosecutorial office in the country to publicly refuse AI-generated police reports and flagged the defense-side argument that defendants could seek information about the underlying AI technology as Brady material.[9] That is FJP’s characterization, not an independently established national ranking. The practical point is more limited and more useful: in King County, the prosecutorial refusal is a known constraint that should be checked before relying on any SPD narrative suspected of AI assistance.

What to verify before relying on an SPD record

The verification path is not complicated, but it is easy to skip if the only visible news is a new chief taking over. A lawyer or risk reviewer should first identify whether the record at issue is a report narrative, force entry, investigative summary, communications statement, or public-facing explanation. The King County memo is most direct for police report narratives. The OPA ChatGPT case is most direct for Blue Team and force-reporting entries. The management action is broader because it addresses the absence of SPD policy guidance on AI use.

• Check OPA Management Action 2024OPA-0333 for the policy gap OPA identified and the recommendation that SPD develop an AI policy.
• Check the city policy-recommendations table for 2025COMP-0007, “Use of Artificial Intelligence,” and note the document cutoff before treating the listed status as current.
• Check SPD’s July 18, 2025 response to the management action for the department’s stated handling of the recommendation.
• Check Closed Case 2024OPA-0333 if the record involves force reporting, Blue Team, or a narrative that may have been drafted or revised through ChatGPT.
• Check Closed Case 2024OPA-0178 if the question is broader internal discipline history involving AI use.
• Check the Sept. 20, 2024 King County Prosecuting Attorney memo before filing or relying on any AI-assisted police narrative in a King County matter.

The question is not only whether an officer used a prohibited tool. Until the department’s formal policy is adopted and visible in city records, the harder questions are attribution, retention, audit trail, and error responsibility. Who authored the sentence? Was the AI input retained? Was the output retained? Was the text edited after generation? Did the officer certify the narrative as personal knowledge? If a name, sequence, or observation is wrong, can anyone tell whether the error came from the officer, the AI system, or a later edit?

Communications records are a weaker but related signal

AI-use questions have also reached SPD communications and public-records practice. PubliCola reported in September 2025 on an anonymous OPA-referred supervisor action alleging that SPD communications statements were created with generative AI without attribution. The article also reported that GPTZero flagged an August 2025 gun-violence post and an April 2025 Barnes statement as likely “100 percent AI,” while noting the limits of AI detectors and reporting SPD counsel’s position that grammar and clarity tools were not “generative” AI under the city policy.[10]

Those detector results are probabilistic, not proof of authorship. The more concrete litigation signal is that PubliCola filed a public-records request for AI inputs and outputs from SPD communications staff.[10] That is the kind of request that can reappear in discovery when parties suspect a public statement, report summary, or official explanation was shaped by AI but not attributed.

Sayles’s appointment gives SPD a new interim chief, not a clean evidentiary slate. Until SPD’s formal AI-use policy is adopted and verifiable in city records, AI-influenced SPD materials should be treated as carrying unresolved verification risk. The safest record review starts with the OPA management action, the city policy table, SPD’s response, the closed-case summaries, and the King County prosecutor memo before anyone relies on the evidence record.

References

1. Seattle Mayor Katie Wilson Names Andre Sayles Interim Chief, SPD Blotter, July 30, 2026
2. Who is Andre Sayles, the new interim Seattle Police Chief?, FOX 13 Seattle
3. OPA MAR 2024OPA-0333 PDF, Seattle Office of Police Accountability, April 4, 2025
4. Policy Recommendations status table, Seattle Office of Police Accountability
5. SPD MAR Response PDF, Seattle Police Department, July 18, 2025
6. Closed Case Summary 2024OPA-0333 PDF, Seattle Office of Police Accountability, February 21, 2025
7. Closed Case Summary 2024OPA-0178 PDF, Seattle Office of Police Accountability, November 8, 2024
8. King County Prosecuting Attorney memo PDF, King County Prosecuting Attorney's Office, September 20, 2024
9. AI-Generated Police Reports: High-Tech, Low Accuracy, Big Risks, Fair and Just Prosecution, June 2025
10. Police Department Acknowledges Using AI, But Says It Isn’t “Substantive” Enough to Label, PubliCola, September 16, 2025