Video footage of federal agents’ killing of Minnesota residents Renee Good and Alex Pretti has galvanized public conversation about the role of immigration enforcement, rule of law and accountability.
Commentators, including Poynter’s Kelly McBride, have argued that this moment in history will be defined by what we choose to record. In Washington, Democratic lawmakers are pressing for systemic adoption of body cameras by federal immigration agents, a request that Secretary of Homeland Security Kristi Noem said she is already moving to adopt.
But the proliferation of footage — from bystanders in the case of Good and Pretti to the widespread adoption of body cameras — creates challenges.
As a two-time Peabody Award-winning broadcast journalist and the founding director of the University of Colorado Boulder’s Visual Evidence Lab — and as fellows at Stanford University’s Center for Advanced Study in the Behavioral Sciences — we understand how video can be easily manipulated. Even when genuine, it can distort reality by triggering our own biases.
Video makes us believe we are witnessing events firsthand, but we are not. Whether you are a journalist, scholar or simply someone watching the news, here are six tips to parsing the value of the video you’re watching.
1. Beware of body camera bias
When we watch body camera video, the lens is pointed away from the officer’s face and body, meaning they are typically out of view. This angle has a psychological impact on the viewer.
Research has found that people who watch an incident recorded by a body camera are less likely to believe the officer acted intentionally than those who watch the same incident recorded by the officer’s car dashboard camera. This is because it’s harder for our minds to process who is at fault when we can’t see the officer. Scholars have also discovered that when even a part of the officer — an arm for example — appears in the shot, it can make a difference. The wider the frame, the fairer the look.
Racial bias is also a factor. Lab experiments show that in police use of force incidents, those who watch body camera footage are more likely to view dark‑skinned individuals as culpable than light-skinned ones — and thus more likely to conclude that the officer’s decision was appropriate and justified. Racial bias can be present in any circumstance, but these experiments found our biases are more pronounced when we watch body camera video than other types of footage. This is another reason to view an incident from as many perspectives as possible.
2. Footage can be manipulated
The Trump administration has used generative artificial intelligence to manipulate footage. In January, the White House posted a digitally altered image of the arrest of Nekima Levy Armstrong, a civil rights attorney, to add tears and darken her skin. In the Department of Homeland Security’s legally required AI use case inventory, the agency disclosed it uses Adobe Firefly, Google Veo 3 and other publicly available digital tools to create and manipulate video content.
More government-controlled footage means more opportunities for video manipulation. In the Armstrong case, the DHS and White House shared two different images of the arrest — one authentic, and another altered by AI.
Last October, Washington Post reporter Drew Harwell used OpenAI to create this deepfake of “body cam footage of cops arresting a dark skinned man in a department store” to draw attention to this problem. Technology has only advanced since then. People are working on solutions to authenticate footage, but these are still in development.
Any piece of media requires critical thinking. We need to ask questions of images just like we do of any other source. What does the image show? Can we corroborate it with other sources? Who shot it, where, and how? Are there other angles?
Be particularly skeptical of videos that are extremely short and may have distorted perspectives — such as an impossible angle. Generative AI can’t unmask ICE agents and reveal their identity. Check whether any news media has reported on the incident. Anyone viewing video, including purported body camera video, should adopt a trust-but-verify framework.
3. Remember: AI makes mistakes
By now, we are all familiar with the idea that AI can “hallucinate” and that these hallucinations have led to false citations in everything from legal briefs to scientific papers. Law enforcement increasingly uses generative AI to transcribe dialogue in police camera footage and draft police reports from it.
Unfortunately, AI makes mistakes. In one instance, a tool wrongly claimed a Utah officer had turned into a frog because it picked up dialogue from “The Princess and the Frog” playing in the background. This comic error highlights how easily generative AI can distort official records.
4. Fight for timely and complete access to footage
Bystander video is available in real time. In September, 70-year-old Air Force veteran Dana Briggs was knocked to the ground by a masked federal agent as he protested the ICE Broadview detention facility outside Chicago. A video of the incident filmed by a fellow protester went viral. The Justice Department charged Briggs with assault.
Six weeks later, U.S. magistrate judge Gabriel A. Fuentes dismissed the case after reviewing the available body camera footage, which the court had allowed the government to shield from public disclosure. But in the intervening month, lies took hold. For video to produce accountability and justice, the public needs to see it immediately.
The government should not be allowed to conceal what it has in its possession — or be selective in the angles it chooses to share. Body camera footage of the 2018 police killing of Harith Augustus in Chicago was released within 24 hours to support the officers’ account of events. But critical dashboard camera footage was withheld for 13 months. Had it been available earlier, the public would have been able to see the same incident from a different perspective, which would have allowed a fuller understanding of the case while it dominated the headlines.
5. Protect your mental health
Repeated viewing of graphic digital content can harm your mental health — even when you didn’t witness events firsthand. A 2015 study of 209 journalists and human rights workers — who were repeatedly exposed to smartphone video and images of war and conflict — found 40% experienced high or very high “personal adverse effects,” such as changes in sleep patterns and harm to their personal relationships, while 20% reported “professional adverse effects,” such as having to take long-term sick leave or quitting their jobs.
There are a number of steps we can take to reduce the risk, the first of which is viewing any video with purpose and intention. The act of journalism, combining the video with other sources of information to tell a story, can be a protective factor in preventing post-traumatic stress disorder. It shows agency and gives us the opportunity to process the footage by creating a narrative, both of which help build resilience over the long term.
Be aware that ruminating on the video — watching it over and over again in an effort to make sense of it — can be psychologically harmful. Basic digital hygiene is important. Monitor screen time; avoid looking at graphic images before bed. Turn off your phone and be with other people — rather than falling inside the video.
6. Shoe leather still matters
Video provides powerful evidence. It can help us understand events we haven’t witnessed ourselves. But it’s just one source of information. It is not a substitute for traditional reporting. Interviewing eyewitnesses and speaking with sources with multiple perspectives on the same event will always be important — especially in an age where any footage is prone to manipulation.
Even in the case of Pretti, where video footage undercut the government’s false claims that Pretti was a domestic terrorist, it took old-fashioned reporting to reveal the truth. The New York Times, The Wall Street Journal and other outlets spent two days examining multiple camera angles and interviewing witnesses to determine that Pretti was shot 10 times after he was already subdued on the ground, his firearm confiscated.
Not every incident will offer the mountain of video and number of onlookers as Pretti’s killing, but the same rules apply. Consider all the available sources. Talk to people. Don’t jump to conclusions. The first answer isn’t always the correct one.
