IE 11 Not Supported

For optimal browsing, we recommend Chrome, Firefox or Safari browsers.

San Antonio Mulls AI Surveillance Camera Pilot Downtown

City officials may seek at least $200,000 for the San Antonio Police Department in the next budget, to add roughly 10 security cameras with artificial intelligence as part of a one-year initiative.

An aerial view of downtown San Antonio, Texas.
Shutterstock
(TNS) — Against a backdrop of recent violent fatalities in downtown San Antonio, a private-sector push for security cameras with artificial intelligence capabilities has won the support of Police Chief William McManus.

The details are still being finalized, but city officials are considering asking for at least $200,000 in the San Antonio Police Department's next fiscal year budget for about 10 security cameras that will be installed as part of a one-year pilot program.

They haven't decided whether the surveillance cameras will use facial recognition technology, which is controversial — but McManus says he's skeptical.

SAPD is frequently approached with proposals from companies that sell AI tools for crime-fighting. But department officials have largely shrugged off the sales pitches — until now.

"These cameras can be programmed these days to alert to an unusual occurrence, to things that may prompt a police response," McManus said. "A fight, a gun, a large crowd getting ready to fight — we would benefit from cameras that could detect all of those things."

The discussions, which began last September, accelerated with pressure from downtown business owners, many of whom are anxious about what they perceive as rising crime, despite data to the contrary. Among their leaders is Trish DeBerry, CEO of Centro San Antonio, a nonprofit that focuses on revitalizing downtown and supporting local businesses.

Since April, there have been a handful of high-profile killings in the downtown area, starting with a shootout at Market Square during Fiesta that left two suspects dead — one killed by police, one by the other gunman — and five bystanders injured.

In May, a stray bullet killed Heidi Silkworth of the Dallas area, an 80-year-old grandmother who had just left a high school graduation at the Alamodome after attending.

On July 22, a high school coach visiting San Antonio for a high school coaches conference, was struck and killed by a stray bullet while sitting on Smoke Skybar's patio.

On July 23, Thomas Darrell Weaver Jr., known as "TJ," a father from Spring, Texas, who was in town for business, died from blunt force trauma injuries to his head sustained during a July 14 assault in front of the Grand Hyatt.

In response, McManus has directed police officers downtown to be more visible. But McManus also believes AI technology could help by sending quick alerts of possible crimes to officers, speeding up investigations and deterring criminals — at least the ones who notice security cameras are watching them.

Deputy City Manager María Villagómez, who oversees the police department, also said the cameras could be beneficial.

"Public safety is important, and we have seen an interest from residents in downtown (for SAPD) to be able to do more," Villagómez said. "This could give us another tool in our tool box to provide that protection to the community."

DeBerry believes enabling facial recognition would boost the results.

Facial recognition technology maps facial features in an image or video, comparing those features with an existing image to determine if the face belongs to the same person — similar to the way iPhone's Face ID works. It also can be used to search a database of images for a match.

"We've got a lot of people who have warrants out for their arrests. We've seen it time and time again where you've got a repeat offender who's not incarcerated and is out perpetrating even more heinous crimes," DeBerry said. "If we had those cameras, we'd be able to identify a person before they commit a really heinous crime — like a murder."

Both McManus and DeBerry say the cameras would be set up in a couple blocks of downtown. DeBerry said it may be wishful thinking, but she'd love to have it up in running in time for Fiesta in April.

"During Fiesta, we had around 25 cars that were broken into in a valet parking lot," DeBerry said. "I can't say that cameras would have completely prevented that from happening. But had we had a security network, and somebody monitoring those cameras could have seen folks gathering and it looked like they were breaking into a car, we could have saved maybe 15 cars."

‘VERY IMPRESSIVE’



DeBerry first got the idea for cameras in September of last year, after taking a trip to Nashville to review the work the city had done to improve safety and cleanliness in their downtown.

Nashville, she learned, had an extensive security camera system.

"I felt like San Antonio really needed something like what Nashville had," DeBerry said. "It was very impressive. They had seen a precipitous decrease in crime."

After DeBerry returned from Nashville, she consulted with downtown businesses and city officials to determine how to proceed. Centro considered a variety options, such as hiring private security guards or off-duty police officers to beef up security.

In the end, DeBerry decided Centro should hire a security company to conduct an audit of the downtown area and determine exactly what was needed. DeBerry contacted two Nashville-based companies she had heard about: Tomahawk Strategic Solutions, which conducts law enforcement training and security assessments, and Herring Technology, a developer of security systems, some of which have artificial intelligence capabilities.

Company representatives traveled to San Antonio in the Spring to conduct the assessment and determine exactly how much a security system would cost. DeBerry said Centro paid about $37,000 for their services.

In the end, Tomahawk and Herring estimated that a security system to help improve cleanliness and safety downtown would cost roughly $1 million, records show.

DeBerry, who meets with McManus every few months, briefed him on the idea.

McManus said the department has not decided which company it wants to hire. Villagómez said they are considering purchasing cameras through Axon Enterprise, the Arizona-based company that the city purchased police body cameras from in 2015.

She said they are also considering working with Fusus, a technology platform recently acquired by Axon that provides crime analytics and develops the artificial intelligence algorithms.

DeBerry said she hopes Tomahawk Strategic Solutions and Herring Technology will be considered, perhaps in some other capacity. She said Centro may help fund the initiative in partnership with the city.

‘A REAL RISK’



SAPD wouldn't be the first agency in Texas to implement cameras with artificial intelligence capabilities. Some even use facial recognition.

The Houston City Council approved a $177,483 contract in June with an AI technology firm to install 64 new cameras throughout the city to improve public safety, though it's not yet clear if the cameras will employ predictive AI algorithms that examine historical data to identify potential risks.

In Dallas, the police department is negotiating a contract with Clearview AI to begin using its facial recognition software, a spokespersons said.

Clearview AI, which claims to be the largest law enforcement database in the world with more than 40 billion photos scraped from the internet, has faced criticism from civil liberty advocates who believe the database violates privacy rights because the photos, though public, were collected without people's consent.

"Facial recognition technologies are a serious threat to communities," said Savannah Kumar, a staff attorney at the ACLU of Texas. "Part of what's so concerning about facial recognition technology is the way that it gives the police such a powerful tool to spy on community members. This includes protesters, political opponents and individuals throughout a community. It causes a real risk of violating people's privacy rights."

Facial recognition is also controversial because it can be inaccurate and wrongly single out minorities. A 2019 study conducted by the National Institute of Standards and Technology that reviewed over 100 facial recognition systems found the algorithms falsely identified individuals who were Asian or Black 10 to 100 times more frequently than individuals who were white.

News reports have documented a handful of cases in which law enforcement arrested the wrong person based on facial recognition.

"It's well documented that certain types of facial recognition technology have trouble differentiating the darker a person's skin is," said Matthew Murrell, a professor at the University of Texas at Austin who studies the intersection of technology and law. "And the problem with that, apart from the racist implications, is that the it just doesn't work as a technology."

Kumar said predictive technologies can be a problem, even if they don't use facial recognition. She said such technologies give police and governments "more and more tools" that can be used in the wrong way.

Kumar cited an example of a blue truck. Technology like the one being considered in San Antonio could be programmed to sort through hours of surveillance footage to identify every time a blue truck is pictured. That could be a handy tool for investigators, who otherwise could spend hours pouring over footage.

But Kumar said that technology could allow the police department to conduct surveillance "at a much higher level" than it otherwise normally does.

"It could mean that just because a particular characteristic about themselves or the property, a person could be ensnared into the criminal legal system," Kumar said. "That can cause a lot of devastating impacts on peoples' lives."

That's why, legal experts say, it's important for agencies to have clear rules in place about how long footage will be stored, who has access to it, and under what circumstances it can be used.

"Agencies should have clear rules about what police can do with these systems," said Peter Salib, an assistant law professor at the University of Houston Law Center. "Are you allowed as an officer to view any footage without some kind of probable cause reason? Are you allowed to go make an arrest based on a system match?"

"I think the answer should usually be no, especially when you can go get corroborating evidence," said Salib, who also serves as an advisor to the Center for AI Safety in San Francisco. "You need limits on how the AI is allowed to be used."

Those guard rails aren't yet in place at the City of San Antonio, though city and police officials said they intend to set them up. Right now, the city has a data security and technology use policy, in addition to a policy that oversees the "Acceptable Use of Generative AI Tools." Generative AI includes tools that create content like ChatGPT. It doesn't include the use of artificial intelligence tools that have pattern recognition or predictive functions.

‘THE PROS AND CONS’



DeBerry said she's familiar with the criticisms of facial recognition. Moving forward, she said it's important for all stakeholders to discuss the pros and cons of the technology.

"There are conversations that have to be had regarding the pros and cons associated with it," DeBerry said. "I would tell you that I believe there are more pros associated than cons."

For his part, McManus said he's very hesitant about the use of facial recognition.

"As we continue the discussions about the cameras and what their capabilities will be, I'm sure that issue may come up, and we'll talk about it," McManus said. "But right now, I understand the sensitivity about facial recognition and the reported pitfalls of it. That's why I've never been hot in the idea ... It's not something I'm anticipating pushing."

Moving forward, Deputy Chief Jesse Salame said the department will prioritize balancing public safety with privacy and civil rights.

"We will establish clear policies to regulate the use and retention of data, ensuring that any technology we deploy adheres to legal standards and best practices," Salame said in an emailed statement. "These cameras are an additional tool to enhance public safety ... AI is not the primary factor in determining when to dispatch an officer based solely on an alert; the final decision will be made by the human monitoring the camera."

Still, Salib says the technology could help. If implemented correctly, cameras with artificial intelligence can improve the efficiency and efficacy of police work. He said some agencies have adopted AI technologies that automatically calculate defendants' risk assessment in determining their bail. That can help remove biases in determining a person's risk to the community, he said.

"In some cases, algorithms do behave better than humans," Salib said. "We should want humans to use the technology to say reduce the number of people in jail."

Aziz Huq, a law professor at the University of Chicago, said it's also important for agencies to vet the company they're hiring. The market for artificial intelligence technology in the criminal justice realm is massive — and growing.

"Many police departments deploy these technologies without clear evidence that the tool is accurate, or without clear evidence that the tool even works," Huq said. "What's the evidence that it actually works?"

(c)2024 the San Antonio Express-News Visit the San Antonio Express-News at www.mysanantonio.com Distributed by Tribune Content Agency, LLC.