Tan v. City of San Jose

Technology-aided record indexing

published federal opinions, CourtListener / RECAP, and the official California Civil Code text). I do not consider every relevant authority. Factual gaps requiring discovery are labeled. Nothing is asserted as established fact unless supported by the record or verified authority.

A. The surveillance architecture

Accepted as true for Rule 12(b)(6) purposes, the Complaint alleges that San José contracted with Flock Safety to deploy automatic license plate readers that grew from a four-camera pilot in 2021 to 149 cameras by the end of 2023 and 474 cameras today, at roughly $2,500 to $2,800 per camera per year. The cameras photograph every passing vehicle; artificial intelligence converts visual features into searchable “Vehicle Fingerprint” fields; and officers can use freeform text search, visual search, Vehicle Journey Maps, convoy analysis, multi-geo search, and related analytic tools. Retention was reduced from one year to thirty days after public criticism. In 2024 the network allegedly generated over 360 million photographs. More than 1,000 SJPD employees have access; data are shared with nearly 300 other Flock customers; and across all users the City’s data were searched nearly 2.5 million times between July 1 and December 9, 2025. The Complaint further alleges strategic placement on residential streets, near immigration lawyers and healthcare facilities, and at city entry-and-exit points, often on both sides of a road or at all four corners of an intersection. Plaintiffs generated a sample of more than 50,000 representative driving routes and allege that the vast majority of captured routes passed two or more cameras—up to fifteen on a single route. The SJPD can also access Flock data from elsewhere in the Bay Area and California, and can fuse ALPR hits with the City’s separate video-camera network. The Complaint alleges that federal agencies obtained “side-door” access through California agencies despite California’s ALPR sharing restrictions.

B. Named-plaintiff injuries and relief

Each named plaintiff alleges repeated photography of his vehicle, conversion into searchable data, thirty-day storage, and warrantless accessibility to more than a thousand officers and hundreds of sharing entities. Tan emphasizes political activism and ICE legal observation. West and Wolfson emphasize residential choke-point cameras and routine tracking. Critically, the Complaint does not allege that any named plaintiff’s plate was actually queried outside this litigation—an omission Defendants treat as dispositive for any challenge to database access. The proposed class comprises all San José residents who were drivers of vehicles photographed after April 15, 2025, or who will be photographed in the future. The requested injunction would require deletion within twenty-four hours absent a warrant or exception, or, alternatively, would bar access to data older than twenty-four hours without a warrant or exception. Nominal damages of one dollar are sought.

A. No “search” under the Fourth Amendment

Defendants argue that ALPRs merely photograph vehicle exteriors on public roads; that United States v. Knotts, New York v. Class, and United States v. Diaz-Castaneda foreclose any reasonable expectation of privacy; that Carpenter v. United States is a narrow cell-site exception that expressly preserves “security cameras”; and that courts in this District—United States v. Rubin and United States v. Butler—together with a nationwide consensus, require dismissal without discovery. Retention of lawfully collected snapshots, they contend, is not independently a search. That framing is rhetorically powerful and doctrinally well-supported. It is also incomplete. The Complaint does not ask the Court to ban roadside photography. It attacks a networked, AI-indexed, multi-jurisdictional, rolling thirty-day movement database with journey maps and association analytics. Defense characterizations that collapse the claim into “static pictures of the outside of vehicles” should be rejected as advocacy, not as a fair restatement of the pleading. 1. Pleading stage versus merits stage. In the Institute for Justice’s parallel Norfolk litigation, Schmidt v. City of Norfolk, the district court denied dismissal on a “whole of movements” theory and later granted summary judgment to the City after discovery. Defendants cite the summary-judgment opinion as if it were a pleading-stage rule of decision. Opposition should force the Court to apply Twombly to San José’s pleaded density and capability facts, not to Norfolk’s post-discovery record. 2. Density and the tipping point. The Schmidt summary-judgment opinion expressly reserves that ALPR systems could become unconstitutional as cameras and capabilities expand—“not today,” but conceivably later. San José’s pleaded facts (474 cameras; multi-capture routes; residential entrance arrays; Bay Area pooling; advanced Flock analytics; video fusion) are engineered to be that tipping-point case. Defense inferences that cameras cover “fewer than 474 discrete locations” should not be resolved against Plaintiffs on a motion to dismiss.

3. Standing bifurcation. Defendants lean on the absence of alleged queries of named plaintiffs’ plates. Parallel cases distinguish Step One (collection and storage) from Step Two (query). Plaintiffs have a stronger standing path on collection and storage. Opposition must bifurcate cleanly and not overclaim a right to enjoin hypothetical queries. 4. The twenty-four-hour remedy is not a merits concession. Defendants argue that seeking deletion after twenty-four hours tacitly admits that capture is lawful. Equitable relief may be narrower than the full scope of a violation. Plaintiffs can maintain that capture-plus-searchable retention is the search while seeking a practical injunction short of dismantling every camera. 5. Non-binding and distinguishable authorities. Judge Bea’s concurrence in United States v. Yang is persuasive only. Sanchez v. Los Angeles Department of Transportation concerned discrete e-scooter trip GPS under a municipal data specification, not a citywide ALPR grid with association analytics. United States v. Porter is out-of-circuit and involved a thin camera network in a criminal suppression posture. Rubin and Butler are important in this District but are not Ninth Circuit holdings. The weight of recent federal decisions favors Defendants on the search question. Plaintiffs’ viable path is to make San José’s density and tooling look categorically different from the thin networks in Porter and Schmidt, and to keep the Court at the pleading stage long enough for discovery on reconstructability.

B. Redundant defendants

Defendants correctly argue that municipal police departments are typically not separately suable when the City is already named, and that an official-capacity suit against the Chief is in substance a suit against the entity. A stipulated dismissal of SJPD, and possibly of Joseph, would sharpen the constitutional fight.

C. Class allegations

Defendants attack the future-looking class as unascertainable and argue that common questions about warrantless querying exceed named plaintiffs’ standing. Rule 12(f) strikes of class allegations are disfavored when issues are better left for Rule 23 practice, but facial overbreadth is a real risk. Plaintiffs should offer a fallback class limited to San José residents whose vehicles were photographed after April 15, 2025, and whose associated data were retained for more than twenty-four hours, and should focus common questions on collection, retention, and access policies rather than individualized queries.

A. Immediate opposition priorities

1. Lead with a corrected claim statement. The Complaint challenges networked searchable retention and analytic reconstruction across a dense fixed camera grid. It does not seek a ban on roadside photographs. 2. Require the Court to accept, under Twombly, the Complaint’s well-pleaded allegations of camera density, multi-capture routes, analytic tools, Bay Area data pooling, and video-camera fusion. 3. Rely on Schmidt’s survival of a motion to dismiss and on its later tipping-point language, while distinguishing the thinner Norfolk record on which summary judgment was later granted. 4. Bifurcate standing. Defend standing to challenge collection and storage. Do not claim standing to challenge database queries unless the pleading alleges that Plaintiffs’ plates were queried. 5. Distinguish Rubin, Butler, the Yang concurrence, Sanchez, Porter, and sister-district dismissals based on thinner networks, different procedural postures, and non-binding status in this Circuit. 6. Preserve discovery. Whether the San José network can reconstruct movements, which Flock features are enabled, how data are shared, and how audits operate are fact questions that should not be resolved against Plaintiffs on the pleadings alone. 7. Offer to dismiss the San José Police Department and consider dismissing Chief Joseph as redundant. Offer a narrowed fallback class definition focused on residents whose vehicles were photographed and whose data were retained for more than twenty-four hours.

B. Evidence and pleading hygiene

Before the October 1 hearing—and ideally before opposition if declarations are used—counsel should obtain and pin the February 9, 2026 ALPR Policy; the Flock Agreement and sharing configuration; Transparency Portal and audit extracts supporting the search-volume and side-door allegations; and the GIS or route study underlying the 50,000-route analysis. The Complaint’s injury paragraphs still refer to retention of “thirty days to one year” while elsewhere stating that current policy is thirty days. That inconsistency should be cleaned in any amended pleading.

C. Defense-win scenarios counsel must plan for

If the Court follows Rubin and Butler and treats San José’s system as legally indistinguishable from thinner ALPR deployments, the Complaint may be dismissed with prejudice on the search question. Even with leave to amend, Plaintiffs may be forced to plead actual queries of named plaintiffs’ plates—which may require audit pulls they do not yet have. Class allegations may be stricken regardless of merits. Those risks are real and should be briefed defensively.