Review Process
The four checkpoints a surveillance or AI tool has to clear in Seattle before deployment: Surveillance Impact Reports, Privacy Impact Assessments, Civil Liberties Assessments, and the Community Surveillance Working Group.
Four Checkpoints, Not One Approval
SMC 14.18 doesn’t ask a department for a single sign-off. It requires a technology to clear four distinct checkpoints, each answering a different question, before it reaches a City Council vote. Leadership evaluating a new tool needs to know all four — because a tool can look strong on one and still fail on another.
| Checkpoint | Question It Answers | Who Produces / Reviews It |
|---|---|---|
| Surveillance Impact Report (SIR) | What does this technology do, what data does it touch, and how is that data handled? | The acquiring department, published for public comment |
| Privacy Impact Assessment | What personal data is collected, retained, or shared, and what are the risks? | City privacy program staff |
| Civil Liberties Assessment | Does deployment create disparate impact or chill constitutionally protected activity? | Civil liberties / equity review staff |
| Community Surveillance Working Group (CSWG) review | What does the affected public actually think, and what conditions do they want attached? | An independent community body, with findings reported to Council |
Surveillance Impact Reports: The Foundational Document
The SIR is the document everything else builds on. It has to describe, in specific and public terms: the technology’s function, the data it collects, how long that data is retained, who it’s shared with (including other agencies), the financial cost, and — critically — a plain-language explanation a non-technical council member and a member of the public can both actually evaluate.
The SIR is published for public comment before Council votes, which means a poorly scoped SIR doesn’t just risk internal pushback — it becomes a public document that advocacy groups, journalists, and residents can and do scrutinize directly.
Privacy and Civil Liberties Assessments: Two Different Lenses on Risk
It’s easy to conflate these, but they ask different questions:
- The Privacy Impact Assessment is a data-handling audit: what’s collected, how it’s secured, how long it’s kept, who can access it. This is the checkpoint most similar to a standard corporate data-privacy review.
- The Civil Liberties Assessment is a use-and-effect audit: even if data handling is technically sound, does how the tool gets used create disparate impact on specific communities, or discourage people from exercising rights — showing up to a protest, a place of worship, a legal clinic — because they know they’re being watched or algorithmically flagged?
A tool can pass the first and fail the second. A camera network with excellent data retention discipline can still raise serious civil liberties concerns if it’s disproportionately sited in specific neighborhoods, or if its output feeds a predictive model that reproduces historical policing bias (a concern covered in depth in Module 3).
The Community Surveillance Working Group
The CSWG is the review step furthest from the department itself: an independent body of community members and civil liberties representatives that reviews SIRs and issues findings and recommendations before Council votes. Its role is not veto power — Council retains the final decision — but its findings become part of the public record Council has to weigh, and its objections are frequently cited in public debate and press coverage of a proposed technology.
Take one AI or surveillance tool your own department is currently considering. Answer, in writing, what a Seattle-style review would require at each checkpoint:
- SIR: Could you currently write a plain-language public description of what data this tool collects and how long it’s retained?
- Privacy Impact Assessment: Who outside the acquiring unit would sign off on the data-handling practices?
- Civil Liberties Assessment: Is there any plausible disparate-impact or chilling-effect concern, even if data handling itself is clean?
- Community review: If an independent community body reviewed this tool publicly today, what’s the single objection you’d expect first?
If you can’t answer one of these confidently, that’s the gap to close before procurement — not after.
A department's Privacy Impact Assessment for a new AI tool comes back clean — data retention is short, access is tightly controlled, and sharing with other agencies is minimal. Command staff conclude the tool is ready for deployment. What does the four-checkpoint model say is missing from that conclusion?
Reveal Answer & Explanation
A clean Privacy Impact Assessment does not substitute for a Civil Liberties Assessment or Community Surveillance Working Group review — those evaluate different risks (disparate impact, chilling effects, public trust) that sound data handling alone does not address.
The four checkpoints are deliberately non-redundant: the Privacy Impact Assessment audits how data is handled, while the Civil Liberties Assessment and CSWG review ask whether the tool’s actual use — regardless of how well the data itself is managed — creates disparate impact or discourages protected activity, and whether the affected public has had a real chance to weigh in. A tool can score well on data hygiene and still generate legitimate civil liberties objections or public opposition that only surface at the later checkpoints.
Up Next
Module 1 has covered why Seattle regulates locally, how that regulation evolved, and what a tool has to pass through before deployment. Module 2 shifts from policy to practice: Operational AI, the first category of tools actually running inside Seattle PD today.