Loading...

As people across the country continue to grieve and call for accountability for the killings of Lorenzo Salgado Araujo and Joan Sebastián Durán Guerrero by ICE agents, Americans find themselves in a ritual that is as predictable as it is grim. Somebody is senselessly killed by law enforcement. Public demands for answers are met with government statements that obfuscate what happened, and refusals to release evidence. Agents’ names are withheld as a matter of policy. Efforts toward formal investigations are obstructed or abandoned entirely. In Minneapolis, it took a lawsuit and more than half a year for state investigators to obtain basic evidence in the killings of Renée Good and Alex Pretti. The sequence is familiar and so is the retreat from culpability.
Much of the coverage about the recent deaths framed this failure as an ICE problem. But the conditions that made these killings not only possible but probable long predate the agency. They are the product of decades of policy, legal, and cultural choices made in response to the deaths of thousands of people at the hands of police officers, sheriffs' deputies, and corrections officers. Choices that have left questions such as whether law enforcement can justifiably shoot a retreating, unarmed person disturbingly unsettled.
Thousands of people are killed by law enforcement in the United States every year. Law enforcement killed 1,318 people in 2025 alone, and more than 16,000 since 2016. Thousands more die in our jails and prisons each year, including at least 663 killed by corrections officers. Most of these killings stay out of public view and the government does not even reliably count them. The police officers or correctional officers responsible for these deaths are rarely prosecuted and even less frequently convicted.
ICE is just the latest agency to avail itself of this system. As killings by immigration agents continue to rise, spend just a minute with us on how we got to the place where state violence is routine and the people who commit it can often expect to walk away.
When decrying ICE violence, critics across the spectrum lament that ICE agents lack the training and standards that govern police officers. It is true ICE has hired at extraordinary speed, slashing training requirements to get agents on the street with a fraction of the preparation that local officers receive. But that does not address the fact that much of police training does more to produce law enforcement violence than prevent it. Police academies continue to devote far more hours to firearms training than to de-escalation or responses to people in crisis. Most police departments require de-escalation only when officers think it’s feasible, and only half of large departments require that officers only use deadly force as a last resort. That police shootings have remained risen through a decade of scrutiny is a direct consequence of this training.
The Supreme Court supplied the legal standard that made it all possible. In 1989, in Graham v. Connor, the Court held that force must be judged not by whether it was necessary but whether it was “objectively reasonable” from the perspective of an officer making a split second decision. What matters is whether the officer’s fear was reasonable, not whether the danger was real. And reasonable doesn’t mean correct. An officer can be fundamentally wrong about a threat, even lethal in that mistake, and remain within the bounds of the law. That logic has spread far beyond the courtroom, shaping department policies including the DHS use of force policy. Remember the ICE officer who claimed he fired on Renée Good’s car because he feared for his life, even though video footage seemed to support claims she was actually driving away?
Even when the law allows for a challenge, systemic conflicts of interest make prosecution unlikely. Local prosecutors rely on police to build every other case in the office and work closely with them on investigations and gathering evidence. Police are an institution district attorneys cannot function without, which makes DAs more likely to serve as a partner to police power rather than a check on it. In 2023, just about 4%of police killings led to an officer being charged with any crime according to Mapping Police Violence.
When the criminal courts fail to hold law enforcement accountable, a civil lawsuit for damages is the last resort. Yet even this path is routinely blocked by qualified immunity. Under this doctrine, an officer cannot be held personally liable unless the violation was "clearly established" as unconstitutional at the time it occurred, meaning a prior court must have already ruled that nearly identical behavior was unlawful. Over decades, the Supreme Court has steadily tightened this standard, demanding an ever-closer factual match between cases before liability can attach. In one case, a Georgia sheriff's deputy tried to shoot a dog, missed, and hit a 10-year-old boy lying beside it. The court granted the deputy qualified immunity because no prior case had this exact set of facts (the shooting of a child while aiming at something else), so the boy's rights were not "clearly established."
Police officers operate with less oversight and transparency than most public sector employees. Over the last few decades, police unions have won protections through collective bargaining and state legislation, effectively sealing police personnel and disciplinary files from the public and making misconduct difficult to uncover. To protect this secrecy, police unions spend tens of millions of dollars lobbying lawmakers and fighting off even basic attempts to open those records. Without access to them, civilians and government officials are left in the dark when it comes to understanding who and how wrongdoing is occurring within police departments and how departments handle incoming complaints and discipline. This lack of transparency has now scaled to the federal level, where even modest efforts to track federal law enforcement misconduct were dismantled last year.
Many of the "reforms” now being proposed for ICE, like mandatory body cameras and retraining, are the same ones we’ve already tried with local police. Adoption has been inconsistent across departments and results have been mixed at best. Federal consent decrees did have some impact where applied but they were too few and too slow and the Justice Department is now actively dismantling that federal oversight. We should be honest about the limits of what these reforms can accomplish as long as the effort to restrain state violence remains a fraction of the effort spent maintaining it, especially as hundreds of billions more move into these enforcement systems in the years ahead.
We can’t afford to scale up, export, or leave unchecked what we’ve already gotten wrong. We’ll be using this note to unpack the faulty thinking about crime, safety, and justice that underpins some of the most consequential discussions and decisions playing out in this American moment.


