Marketing vs. reality

San Mateo County’s Project Sherlock, a $12 million AI policing system developed by C3 AI, has largely failed to deliver promised investigative benefits after three years. A timeline projected 14 agencies to be fully live by January 2025, but seven agencies—Atherton, Belmont, East Palo Alto, Foster City, Menlo Park, San Bruno, and South San Francisco—report lacking a fully functional product. A senior Sheriff’s Office analyst said: ‘My no BS assessment is that we’ve been working with them for two years and they have a barely functional product.’ The system has faced delays, login bugs, and data integration issues.

On February 8, 2024 Flock Safety issued a press release headlined ‘Flock Safety is Instrumental in Solving 10% of Reported Crime in U.S.’ The release analyzed crime-clearance data from 123 law enforcement agencies collected between April and June 2023. In March 2024, 404 Media reported the study was conducted by Flock employees with oversight from two academic researchers, one of whom said ‘I personally would have done things much differently’ and that the police data was ‘too varied and incomplete’ for meaningful statistical analysis. The researcher also said the two were brought into the study late in the process. The findings drew significant criticism from other criminologists.

Voice of San Diego obtained San Diego Police Department data covering 584 ShotSpotter activations from late 2016 through February 2020. SDPD classified 72 activations, 12.3 percent, as unfounded. ShotSpotter’s senior vice president of marketing Sam Klepper said the company had a 0.5 percent false-positive rate. Police made two arrests while responding to activations; nearly 60 percent of dispatches produced no report. SDPD Lt. Jeffrey Peterson wrote in December 2019, ‘I don’t support keeping it based on the cost.’ The city’s $1 million four-year contract was set to expire Nov. 21.