SFPD authorized to kill suspects using robots in draft policy

A policy proposal heading for Board of Supervisors approval next week would explicitly authorize San Francisco police to kill suspects using robots.

The new policy, which defines how the SFPD is allowed to use its military-style weapons, was put together by the police department. Over the past several weeks, it has been scrutinized by supervisors Aaron Peskin, Rafael Mandelman and Connie Chan, who together comprise the Board of Supervisors Rules Committee.

The draft policy faces criticism from advocates for its language on robot force, as well as for excluding hundreds of assault rifles from its inventory of military-style weapons and for not including personnel costs in the price of its weapons.

Peskin, chair of the committee, initially attempted to limit the SFPD’s authority over the department’s robots by inserting the sentence, “Robots shall not be used as a Use of Force against any person.”

The following week, the police struck out his suggestion with a thick red line.

It was replaced by language that codifies the department’s authority to use lethal force via robots: “Robots will only be used as a deadly force option when risk of loss of life to members of the public or officers are imminent and outweigh any other force option available to SFPD.” 

This could mark a legal crossing of the Rubicon for the city: Robot use-of-force has never before been approved, nor has it ever been prohibited, in San Francisco. A version of this draft policy was unanimously accepted by the rules committee last week and will come before the full board on Nov. 29.

“The original policy they submitted was actually silent on whether robots could deploy lethal force,” said Peskin. He added that he decided to approve the SFPD’s caveated guidelines because the department had made the case that “there could be scenarios where deployment of lethal force was the only option.”

Advocates and lawyers who oppose the militarization of the police are less convinced.

“We are living in a dystopian future, where we debate whether the police may use robots to execute citizens without a trial, jury, or judge,” said Tifanei Moyer, senior staff attorney at the Lawyers’ Committee for Civil Rights of the San Francisco Bay Area. Moyer leads the organization’s work on police misconduct and militarization.

“This is not normal,” she wrote over email. “No legal professional or ordinary resident should carry on as if it is normal.”

The SFPD has 17 robots in its arsenal, 12 of which it describes as fully functional. According to police spokesperson Officer Robert Rueca, they have never been used to attack anyone. The robots are remote-controlled, and are typically used to investigate and defuse potential bombs or to surveil areas too awkward or dangerous for officers to access.

Uses defined in the new draft policy include “training and simulations, criminal apprehensions, critical incidents, exigent circumstances, executing a warrant or during suspicious device assessments.”

And, in extreme circumstances, they can be used to kill.

How are robots used lethally?

In 2016, the Dallas police force strapped plastic explosives to a robot and used it to blow up a sharpshooter who had killed five officers, in the first U.S. instance of a police robot killing a suspect. One of the SFPD’s robots, the Remotec F5A, is the same model as the one used by Dallas police.

More recently in Oakland, a policy on lethal robots came before the city’s police department’s civilian oversight council. One device they discussed was the PAN disruptor, a device that can be attached to a remote-controlled robot and uses a blank shotgun shell to disable a bomb by blasting it with pressurized water. Oakland police acknowledged that, in emergencies, they could arm it with live rounds. The SFPD also has multiple PAN disruptors that can be attached to robots and fire shotgun shells.

Last month, Oakland police ultimately backed down and removed language that would have allowed them to kill using robots. They said they hope to pursue the option in the future.

Rueca said that the San Francisco Police Department “does not have any sort of specific plan in place” for how lethal force would be applied with robots as “the unusually dangerous or spontaneous operations where SFPD’s need to deliver deadly force via robot would be a rare and exceptional circumstance.”

Why is this happening now?

Cities across California are currently drafting new policies on the use of military weapons by local police forces, thanks to a state law called AB 481, which passed last year. Figuring out the force options of robots is one small part of the law’s remit.

The law mandates that every police force in California must annually report its stock of all military-style weapons, their cost, how they can be used, and how they were deployed in the prior year. The law gives local authorities — in San Francisco’s case, the Board of Supervisors — the ability to annually reject or accept the rules governing how the weapons are used.

The Board will also be required to sign off on any new military-style equipment before purchase, although the police will be able to replace any existing equipment up to a value of $10 million without approval.

Most advocates opposed to the militarization of the police hail AB 481 as a step in the right direction for accountability and transparency. But concerns have been raised that some jurisdictions have not gone far enough in limiting how military-style weapons can be used.

Jennifer Tu, a fellow with the American Friends Service Committee, has been tracking how police departments across the state are implementing AB 481.

“My suspicion is that most policies will have left room for robots to use force,” said Tu. She said that it was her understanding that most departments have not mentioned robots at all, which means they are subject only to generic restrictions.

The ACLU has published advice on the use of robots by police, and notes that the limited situational awareness of robots, compared to in-person officers, make it more likely that force is “used inappropriately and/or on the wrong targets.”

“There is a really big difference between hurting someone right in front of you, and hurting someone via a video screen,” said Tu.

What else is in the draft policy?

Tu contended that, on top of the issue of robot force, there are other problems with San Francisco’s draft policy, as it currently stands.

In its initial submission, the SFPD omitted all of its 608 semi-automatic assault rifles, 64 machine guns, and 15 submachine guns from the new use-of-force policy. According to Peskin, these were added in when he pushed back on their omission. But in the department’s latest version, which is set to come before the supervisors next week, 375 of the semi-automatic assault rifles are missing again.

The rationale given for the removal of these assault rifles from the policy: The Chief of Police defines them as “standard-issue service weapons.”

Others disagree with that assessment. “We don’t see regular officers walking around with assault rifles,” said Allyssa Victory, staff attorney with the criminal justice program at the ACLU of Northern California (and recent Oakland mayoral hopeful). “Just writing a policy doesn’t make it so.”

Victory added that shotguns and handguns can be omitted because they are standard issue, according to AB 481, but no such exemption applies to assault rifles.

“The law defines ‘military weapons,’ not the chief of police,” wrote civil rights lawyer Moyer over email. “San Francisco is not the only department to attempt to redefine ‘military weapons’ so as to justify hiding their use, costs, and upkeep from the public.”

“If the law defined military weapons as bubble gum, then the police department would have to disclose their use of bubble gum,” she wrote.

Tu added, “I really think it is confusing to the public if we don’t have those assault weapons reported.” Their omission would mean that in future annual reports, the police would not need to declare how the guns had been used or who had been injured by them.

Another point of contention with advocates is that the SFPD has not included personnel training or maintenance times in their valuation of the cost of their military-style weapons. This appears to be required by AB 481, which states that costs must include “acquisition, personnel, training, transportation, maintenance, storage, upgrade, and other ongoing costs” of the weapons.

But the SFPD rejected a suggestion from the American Friends Service Committee to include personnel costs. The department said that maintenance and training occur during normal work hours, and that their human resources management system cannot track different types of work done by officers, so “there is no compelling reason to track in the suggested manner.”

It remains to be seen if the policy, as it stands, will be approved by the Board of Supervisors, and what limitations will ultimately be placed on the police department’s military-style weapons, including its robots. And, once the rules are settled, the process will begin again with the Sheriff’s Department, which will need to create its own policy to stay in compliance with AB 481.

“The great news about this thing is that it can be evolved,” said Peskin, adding that policy must be scrutinized and approved every year if the SFPD wants to keep using its weapons.

“And I think we are starting off in a good place.”

This policy will be discussed at the Board of Supervisors meeting on Tuesday November 29. The meeting starts at 2 p.m. and the police equipment policy is agenda item 28. More details can be found in the full meeting agenda.