Watch the watchers: Atlanta plan calls for alerts on suspicious surveillance searches

Repeated searches involving the same person or car. Unusually heavy access. Activity that does not fit a user’s role. Under Atlanta’s proposed surveillance rules, patterns like these could trigger automated warnings for reviewers—not automatic findings of wrongdoing.⁠[1]

That safeguard is part of ordinance 26-O-1401, listed for Finance/Executive Committee review Oct. 14 at 1:30 p.m. But forwarding depends on a Public Safety and Legal Administration Committee recommendation. The records reviewed do not establish what Public Safety did Oct. 12 or whether the ordinance has been adopted; the latest recorded disposition supplied is Council’s Oct. 5 return of the measure as held.⁠[1][2]

The proposal would require medium- and high-impact surveillance systems to flag potential misuse where technologically applicable. Vendors lacking that capability would have to develop substantially equivalent functionality on a city-set timetable. The text contemplates a 12-month transition for existing vendors to demonstrate the capability or an acceptable plan—but explicitly leaves the final transition period and contractual implementation mechanism as unresolved law and policy questions.⁠[1]

Human review would remain essential. Alerts would have to be reviewed and documented, and an alert alone would not establish a violation. Atlanta Police would conduct monthly audits of medium- and high-impact systems, with material findings, suspected violations and corrective actions included in the chief’s quarterly reporting, subject to legal and confidentiality restrictions. High-impact systems would also receive independent technical audits at least annually; medium-impact audits would be risk-based at the City Auditor’s discretion.⁠[1]

The broader framework would sort technology by capability and risk—not price alone. Atlanta Information Management would classify it: low-risk tools could receive administrative approval, medium-impact tools would face panel review, and high-impact tools would require impact reports, use policies, panel review, public participation and affirmative Council approval before deployment, subject to the proposal’s exceptions. Facial recognition, persistent tracking and automated public-safety profiling are among the high-impact categories.⁠[1]

Existing systems would not face retrospective panel review or Council reapproval simply because they predate the rules. Departments would instead inventory their technology within 90 days after the rules took effect, then complete impact reports and use policies for existing medium- and high-impact systems within 180 days after classification. Those systems would still face ongoing oversight, and material expansions would trigger review.⁠[1]

Temporary emergency and major-event deployments could proceed before advance review was complete, but only for the necessary purpose and duration. Continued or recurring use would require the otherwise applicable review. Separately, vendors would be barred from selling or monetizing city surveillance data, with exceptions for legally required disclosures and sharing strictly necessary for contracted services that a written city agreement expressly authorizes.⁠[1]