Seattle Police Department holds hackathon centered on tech for redacting videos recorded by officers from dashcams and body-worn cameras
Inside the Seattle Police hackathon: A substantial first step — The Seattle Police Department (SPD) held its first-ever hackathon on Friday.
Context & Ripple Effects
In late 2014, Seattle Police Department ran its first-ever hackathon, and the target was one of the most tedious chokepoints in police transparency: manually blurring faces and license plates out of dashcam and body-worn camera footage before it can be released. It fits a pattern at SPD of using lightweight programs instead of procurement — weeks earlier it had quietly stood up an anti-swatting registry letting people who fear being swatted warn officers in advance.
The reason redaction matters is volume: every camera a department deploys multiplies the hours that must be scrubbed before any footage goes public, and in-house IT rarely keeps up.
First-order effects
- SPD gets prototype redaction tooling from civic technologists rather than waiting on vendor contracts, and the department gains a working template for opening its engineering problems to outside developers.
- Departments sitting on unreleasable footage backlogs get a demonstration that the bottleneck can be attacked with software rather than staff hours.
Second-order effects
- If manual scrubbing stays too slow, the market answer is automation — by 2024 US departments are running AI from Truleo and others over millions of hours of unreviewed body-cam footage, exactly the volume problem this hackathon was chipping at.
- Widening access to police video tools cuts both ways: some departments already run facial recognition searches on behalf of other jurisdictions, so better tooling built for transparency can be repurposed for surveillance that evades local laws.
Third-order effects
- The set of footage police must process keeps growing beyond their own cameras — San Francisco training documents already treat video from continuously recording autonomous vehicles as obtainable evidence — making redaction and chain-of-custody tooling permanent infrastructure rather than a one-off project.
- As police tech becomes outsourced software, security of the pipeline itself becomes the attack surface; the ODIN breach showed what leaks when vendor servers hold raid plans and credentials, and researchers had earlier shown body cams themselves can be tracked and tampered with.
The trend: Police video is shifting from a manual evidence-handling chore to automated, vendor-supplied pipelines — and the governance questions about who builds, accesses, and secures those pipelines start with moments like this hackathon.