Flock's AI identifies drivers by movement, not plates
Flock Safety spent years telling the public its cameras only read license plates. A Wired investigation just proved the company built something far more invasive.
Flock Safety has built an AI-powered surveillance system called OS Investigate that can identify drivers and track their movements without ever reading a license plate. Wired reconstructed code from Flock's own login pages and found a tool already in testing with some police departments that flips the company's core promise on its head. Where Flock previously marketed its cameras as plate-reading systems that "cannot recognize, identify, or track individuals," OS Investigate does exactly that — using patterns of vehicle movement to surface names, home addresses, relatives, and associates. The system accesses police case files, 911 dispatch logs, and commercial databases containing Social Security numbers, dates of birth, and phone numbers, turning a single plate into a full background check with one command.
The system ships with 69 prewritten prompts that officers can select or customize. One asks the AI to find witnesses based on vehicles most frequently seen in a neighborhood over the past two weeks. Another instructs the system to list everyone arrested more than twice in two years, map where they live, and "do a workup on the top 3 individuals." Nineteen of the prompts describe hunting for behavioral patterns rather than looking up specific records: vehicles that visited three or more retail locations in three days, or multiple banks in a week, or gas stations between midnight and 5 am. A built-in filter strips out buses, semi trucks, and work vans so only ordinary drivers appear in results.
OS Investigate can also identify a target's "associates" by counting how often other vehicles appear at the same cameras within a two-minute window, ranking them by a confidence threshold defaulting to 0.75. Give it one plate and it returns up to 20 associated vehicles, which it then converts into names and addresses. Jay Stanley, a senior policy analyst at the ACLU, compared the system's scope to China's surveillance infrastructure, saying there is "very little space between what can be done and what is being done." Flock says the product is still in development and capabilities may change before broader release, but the code is already live on the company's own servers.
Salesforce opens 200-plus APIs to AI agents through MCP
Salesforce expanded its Headless Data 360 platform to support the Model Context Protocol, giving AI agents direct access to more than 200 Salesforce APIs without requiring developers to build custom integrations. The move means agents from Claude, Cursor, ChatGPT, or Salesforce's own Agentforce can query customer data, run transformations, and generate insights using natural language — for example, asking for "lifetime value of all customers for electronics purchases, excluding software" and having the agent construct the query, build the semantic model, and execute it automatically. Salesforce described the original Headless 360, released in April, as the "front door" to the ecosystem; this expansion goes deeper, providing data intelligence and governance directly to agents. The MCP server is Salesforce-hosted and includes prepackaged Skills for common tasks like data modeling and field mapping, with custom Skills arriving later in August. We explained the Model Context Protocol in our primer — AI 101 — What is MCP?.
Should police departments need a warrant before deploying AI tools that can identify anyone by how they drive? Tell us in the comments.
