A cop searched 19,000 Flock cameras to type "LMAO"
The pitch for automated license-plate cameras is that they catch stolen cars and wanted suspects. Flock Safety's own audit logs suggest a lot of the searches were for something else entirely.
A review of Flock Safety's search logs found officers using the one field meant to justify a search to type jokes and gibberish. In July 2025, an officer with the Lake County, Indiana, sheriff's department ran a plate search across more than 19,000 cameras covering 1,558 cities and towns. The stated reason, as recorded in the log: "LMAO."
The pattern is wider than one officer. In records spanning 2023 to late 2025, the Electronic Frontier Foundation found search reasons including "LOL," "Hehe," "Haha," "idk," "blah" and strings of keyboard mash such as "asdfg." One officer entered: "robbery I don't remember the case number leave me alone." More than 6,300 searches across 30-plus agencies simply said "TBD."
Why it matters: the reason field is the only accountability artifact a plate-camera search leaves behind, and Flock spent years telling departments and city councils that it was how misuse would be caught. In late 2025 the company replaced the free-text field with a list of pre-populated categories. EFF called that change a loss for transparency — a generic category can say less about a search's purpose than a joke does. Agencies contacted by EFF said officers had been counselled or that the searches would be reviewed, and the company has been tightening rules all year under a widening surveillance backlash.
This thread keeps repeating. We covered the retention cut after the last abuse reports — Flock slashes plate-camera data retention to 7 days after abuse reports — and the capability expansion that followed — Flock's AI identifies drivers by movement, not plates. Both changes came after exposés, not after internal audits. That is the tell: the oversight is reactive, so the logs keep producing the same finding.
Cisco rebuilt Splunk around agents, and put a meter next to them. At its .conf user event on Tuesday, Cisco's Splunk unit rolled out a platform overhaul aimed at the "agentic enterprise": Cisco Data Fabric, which searches data where it already lives — cloud object stores, data lakes, Snowflake, Databricks — instead of forcing customers to ingest everything into Splunk, directly attacking the ingestion bill that has always been the platform's weak point. The company also expanded its Nvidia partnership into the Cisco AI POD, which runs Splunk AI on-premises, in private clouds and in air-gapped environments on Nvidia accelerators, with support for self-hosted models. A new multi-year agreement with AWS turns a long integration relationship into joint product development of agent-specific security tooling.
Two pieces are worth watching for anyone already running agents in production. Splunk Agent Observability watches the infrastructure, models, orchestration and memory behind an agent, evaluates every event without sampling and can block an action at runtime if the agent steps outside its boundaries. The new Tokenomics capability tracks token consumption across models and coding assistants, attributes the bill to teams and users, and forecasts consumption before a billing period closes. Pricing an agent fleet is currently guesswork for most enterprises; putting it in the same console that watches the agent is the more defensible half of the launch.
Meta put a list price on its AI and its creator perks. The company introduced Meta One, a subscription family that starts at $7.99 per month and bundles individual app subscriptions with things users previously got free or not at all: 48-hour Instagram stories, a bold follow button, a verified badge, higher search ranking and monitoring for impersonation accounts. Creator and business tiers run from $14.99 to $499 per month. Meta's AI features ride along inside those tiers rather than being sold as a standalone product — the same bundle-everything play it ran with iCloud-style subscriptions elsewhere in the market, and a quiet signal that consumer AI is being monetized as a perk, not a destination.
What to watch: whether any of the departments named in the Flock review face consequences beyond a counselling note, and whether Flock's category-based reason field produces logs anyone can audit.
Should search logs from AI camera networks be public by default? Tell us in the comments.
