Washington State Requirements — The July 2026 Deadline
Washington State has a specific AI accountability mandate for public agencies with a hard deadline. Here's what the Task Force requires and what it means for your department.
A State-Level Line in the Sand
CJIS and HIPAA are federal frameworks. They set a floor — baseline requirements that apply to every law enforcement agency and EMS provider in the country.
Washington State has gone further.
In 2024, Washington established the Joint Legislative Task Force on the Use of Artificial Intelligence in the Criminal Legal System. The Task Force was created in direct response to concerns about AI tools being deployed in criminal justice contexts — predictive policing, evidence processing, report writing — without consistent accountability standards.
The Task Force has a mandate: by July 2026, Washington public agencies using AI in criminal justice contexts must meet specific requirements for transparency, oversight, and accountability.
This is not a distant policy discussion. It’s a requirement with a deadline, and if your agency uses AI tools in law enforcement or criminal justice, you are in scope.
What the Washington AI Task Force Is Focused On
The Task Force was charged with examining four specific areas:
1. Transparency About AI Use
Agencies must be able to say — clearly and specifically — what AI tools they are using, for what purpose, and how those tools make or influence decisions.
“We use some AI tools” is not sufficient. The requirement is specificity: which tools, in which workflows, with what documented decision-making role.
2. Bias Auditing
AI tools used in criminal justice contexts must be audited for bias — specifically, whether they produce disparate outcomes for individuals based on race, ethnicity, gender, or geography.
This includes predictive patrol tools, risk assessment instruments, and facial recognition systems. The bias audit must be documented and reviewable.
3. Human Oversight Requirements
For any AI tool that produces a recommendation affecting a criminal justice decision, the requirement is that a qualified human must review and take accountability for that decision.
This formalizes at the state level what the Traffic Light Framework teaches operationally: AI advises, humans decide.
4. Accountability for AI-Assisted Decisions
When an AI tool contributes to a decision that results in arrest, detention, prosecution, or sentencing, that contribution must be documented and disclosable. Defense attorneys must be able to access information about AI tools used in their client’s case.
This is the Brady implication formalized in state policy: if an AI tool played a role in a criminal case, that’s discoverable information.
Why July 2026
The Task Force timeline reflects the pace of AI deployment in Washington State agencies. Multiple departments were already piloting or deploying AI tools when the Task Force was convened — the deadline was designed to give agencies time to inventory their tools and build compliance frameworks before mandatory enforcement begins.
For agencies that haven’t started: the time is now. An inventory of current AI tools, documented policies, and bias audit processes takes months to build correctly. Waiting until June 2026 is not a viable approach.
What Agencies Must Have by July 2026
Based on the Task Force mandate, Washington public safety agencies will need to document:
| Requirement | What It Means in Practice |
|---|---|
| AI tool inventory | A complete list of every AI tool in use, its vendor, its purpose, and its data inputs |
| Use policy for each tool | Written guidance on how the tool is used, what decisions it informs, and what human review is required |
| Bias audit documentation | Evidence that each tool has been evaluated for disparate impact |
| Training records | Documentation that personnel using AI tools have been trained on their limitations and proper use |
| Disclosure process | A mechanism for providing AI tool information in response to discovery requests |
How This Affects You on the Ground
If you work for a Washington State public safety agency:
You may be asked to document your AI use. Incident reports, investigation files, and case documentation may increasingly need to include notation of any AI assistance used.
Training requirements will increase. Completing this course is a step toward the training documentation requirement. Agencies will likely need to track AI training for all personnel using AI tools.
Vendor relationships will change. Agencies will need to pressure vendors for bias audit data, documentation, and compliance certificates. Vendors that can’t provide this documentation won’t survive agency procurement reviews post-2026.
Discovery obligations are real. If you used an AI tool in a case — to draft a report, analyze evidence, flag a suspect, or recommend a patrol location — that use may be disclosable to defense counsel. This isn’t a reason to avoid AI. It’s a reason to use it properly and document it.
Use these questions to assess where your agency stands relative to the July 2026 requirements.
AI Tool Inventory
- Does your agency have a complete list of every AI tool in active use?
- Does each tool have a documented purpose and data input description?
- Are you aware of any AI features embedded in software your agency uses that haven’t been formally inventoried?
Use Policies
- Is there a written policy for each AI tool specifying what decisions it can and can’t inform?
- Do officers know what review is required before acting on AI output?
- Is there a clear process for documenting AI assistance in case files?
Bias Auditing
- Has your agency requested bias audit data from AI vendors?
- For predictive tools, has the agency reviewed whether output patterns show geographic or demographic concentration?
Training
- Have all personnel using AI tools completed documented training?
- Is there a process for ongoing training as tools change or new tools are added?
Disclosure Process
- Does your agency have a process for responding to discovery requests that ask about AI tool use?
- Are prosecutors in your jurisdiction aware of AI disclosure obligations?
If you answered “no” or “I don’t know” to more than two or three of these, your agency has work to do before July 2026.
The July 2026 requirements are not purely a technology compliance issue. The bias auditing requirement, the use policy requirement, and the training documentation requirement are operational and administrative — they require action from patrol commanders, EMS supervisors, records managers, and training officers, not just IT staff.
IT may be able to inventory the tools. They cannot write use policies for how patrol officers engage with AI dispatch tools, or document what training dispatchers have received on AI triage tools.
The compliance burden is distributed across the agency. Waiting for IT to handle it is how agencies miss the deadline.
A defense attorney requests all documentation related to AI tool use in the investigation of their client. The detective says the AI tool was 'just used for a quick search' and didn't affect the outcome. What is the correct approach?
Reveal Answer & Explanation
The detective’s characterization doesn’t determine discoverability — the legal standard does.
Under Brady v. Maryland and Washington’s AI accountability framework, the relevant question is not whether the AI “affected the outcome” in the detective’s estimation. The question is whether the use of an AI tool is material information that defense counsel is entitled to.
If an AI tool was used at any point in the investigation — to process evidence, flag a suspect, analyze data, draft documents — that use should be documented and producible in response to a discovery request.
Practically, the correct approach:
- Document AI tool use at the time it occurs — in the case file, not in memory
- When a discovery request arrives, review the documentation and produce it with other case materials
- Prosecutors should review AI tool use records as part of their Brady review process
- Do not attempt to characterize whether the AI “mattered” — that determination belongs to the court and defense counsel
The agencies that build AI documentation into their standard case file practices now will be far better positioned when discovery requests arrive post-2026 — because they’ll have the records to produce.
Up Next
The final lesson of this course brings everything together — and returns to the most important question in AI for first responders: who is ultimately responsible?
Lesson 4: You Are the Last Line of Defense.