SIFT with WikiPedia and AI: Killings in Minnesota

an hour ago 9

The national and international events of the past several weeks have been positively dizzying and even disorienting. As I write this on January 27, 2026, we (as a nation) are recovering from the shock of two killings by federal law enforcement officials (ICE and CBP) in Minneapolis, Minnesota. Incredibly, even though both shootings / killings […]

The national and international events of the past several weeks have been positively dizzying and even disorienting. As I write this on January 27, 2026, we (as a nation) are recovering from the shock of two killings by federal law enforcement officials (ICE and CBP) in Minneapolis, Minnesota. Incredibly, even though both shootings / killings by law enforcement were well documented by witness videos, completely different narratives of events have been presented and amplified by representatives of the US government and by conservative organizations like Fox and conservative social media influencers.

A composite image titled "Divergent truths" contrasts different media portrayals of a news story involving the death of a man named Alex Pretti. On the left, a Fox News headline blames a "radical network" for putting Pretti in harm's way, paired with a social media meme featuring a photo of politicians and text claiming violence would not have occurred if Democrats hadn't encouraged the obstruction of federal law. On the right, an Al Jazeera article titled "Who was Alex Pretti?" features an outdoor photo of him smiling and includes a subheadline noting that witnesses refute government claims that he brandished a gun before being shot by federal agents in Minneapolis.
Divergent Truths (CC BY 4.0) by Wesley Fryer

WikiPedia can be a helpful source of information for breaking news and controversial events like these. Unlike mainstream and social media “news” sources and influencers, who can present divergent perspectives and even contradictory “facts,” WikiPedia presents the SAME PAGE of information for each article to viewers, but maintains TALK PAGES for each article where controversies and disagreements are hashed out among article editors and site admins. In the case of these recent killings, I recommend checking out BOTH the articles and the TALK PAGES for:

Each semester in my middle school media literacy classes, I share a lesson I call, “LaunchPad WikiPedia,” in which students learn how to use a variety of features of WikiPedia to not only understand different topics and issues, but also how disagreements about them are “hashed out” among users and admins. We create a sketchnote of John Green‘s outstanding video from 2019, Using Wikipedia: Crash Course Navigating Digital Information #5 (13:30).

One of the key media literacy skills I introduce to students through our “Brain Hacking InfoPics” lesson series as well as “Froot Loop Conspiracy Theories” unit is SIFT. Developed by Mike Caufield at the University of Washington’s Center for an Informed Public, SIFT is a media literacy strategy that encourages people to STOP when they encounter potentially emotional / polarizing information or an unfamiliar news source. SIFT nexts directs people to INVESTIGATE the source, FIND trusted coverage, and lastly TRACE claims to their original source.

This infographic outlines the SIFT method for media literacy using four icons and text labels arranged in a horizontal sequence. The process begins with S for "Stop," represented by a hand inside an octagonal stop sign. This is followed by I for "Investigate the source," paired with a magnifying glass icon. Next is F for "Find better coverage," shown with a checkmark. Finally, T stands for "Trace claims, quotes and media to the original context," illustrated by a branching flowchart icon.
Caufield, Mike, “SIFT (The Four Moves).” 19 June 2019, https://hapgood.us/2019/06/19/sift-the-four-moves/.

Since SIFT was originally developed and John Green recorded his outstanding video on using WikiPedia as a launchpad for learning, artificial intelligence (AI) platforms have entered our information landscape and are increasingly providing direct “answers” to user questions, instead of a series of web links like a Google search. Given their near-universal tendency to confabulate / hallucinate, AI platforms present significant challenges.

AI tools also offer incredibly powerful opportunities for deep information analysis and fact checking, however, and Mike Caufield’s latest “Deep Research” AI superprompt strategy, which he calls, “SIFT for AI,” is a transformational tool for our personal media literacy toolkits. I facilitated a webinar for the Media Education Lab in November 2025 exploring this strategy, titled “Fact Checking with AI Superprompts.”

This past weekend, as we prepared for a significant winter ice storm in the southeastern United States, I shared a post on my “Resist and Heal” Substack that I cross-posted on my personal Facebook page titled, “Death Squads in the USA.” That post sparked a flurry of thoughtful comments (35 to date) from the politically diverse group of folks I’m connected to via that Facebook profile.

One of those commenters, a college classmate of mine, shared a long response that included more than ten different claims about the killings of both Renee Good and Alex Pretti, as well as past actions by other state and federal officials. A Facebook comment thread seemed like a VERY inadequate medium to thoughtfully respond, so I used Caufield’s “Deep Research” AI superprompt with Claude Pro AI to investigate the claims. The 7.5 pages of results are both helpful and eye opening, and provide a timely example of how AI superprompts like this can be extremely useful when we find ourselves confronted with an overwhelming volume of contradictory and controversial claims about current events.

In addition to offering a “deep dive” fact check into the presented claims, Caufield’s “Deep Research” superprompt also offers a variety of “commands” which can provide further insight into a given topic or claim set. I opted to choose, “explain like I’m in high school,” and the following are the unedited Claude Pro AI results. (You can also view an archive of my entire conversation with Claude – which includes everything except my friend’s original FB comment.)


The Basic Situation

You’ve probably seen posts online about ICE (Immigration and Customs Enforcement) agents shooting people in Minneapolis. There’s a lot of conflicting information flying around, so let me break down what we actually know.


Two Separate Shootings, Two Different People

Shooting #1: Renee Good (January 7, 2026)

Renee Good was a 37-year-old mom of three who was part of a neighborhood group that watched ICE agents to alert immigrants when raids were happening (kind of like a neighborhood watch, but for immigration enforcement).

On the morning of January 7th, she was sitting in her parked SUV in the middle of a residential street, essentially blocking traffic to slow down ICE vehicles. An ICE agent named Jonathan Ross approached her car. What happened next is where the stories diverge:

  • What the government says: She tried to run over the agent with her car, so he shot her in self-defense. They called it “domestic terrorism.”
  • What the videos show: Multiple news organizations did frame-by-frame analysis of bystander videos. They found that her steering wheel was turned away from the agent right before he fired. The Minneapolis mayor watched the videos and said the government’s story was “bullshit.”

She was shot three times and died.


Shooting #2: Alex Pretti (January 24, 2026)

Alex Pretti was a 37-year-old ICU nurse who worked at a VA hospital taking care of veterans. He was also legally allowed to carry a gun (he had a permit).

On January 24th, he was in an area where ICE was conducting arrests. He saw agents push a woman to the ground and stepped in to help her up. Then:

  • What the government says: He approached agents with a gun, refused to be disarmed, and “violently resisted.” They said he came there to “inflict maximum damage” and called him a “domestic terrorist.”
  • What the videos show: He was holding his cell phone, not a gun. Video shows an agent reaching into the scuffle and taking Pretti’s gun from his holster. Less than a second later, other agents shot Pretti 10 times. He never fired his weapon — he didn’t even have it in his hands.

The Original Post’s Claims — What Checks Out and What Doesn’t

The post you saw makes a lot of claims. Here’s the scorecard:

WRONG: “Body cam footage shows her make eye contact, smile, and gun the engine”

The footage released wasn’t even from a body camera — it was from the agent’s personal cell phone. And it actually shows her saying “I’m not mad at you” while sitting still. Video analysis shows her turning the wheel away from the agent before he shot.

WRONG: “She dragged one cop”

There’s no video evidence of this. The mayor said the agent walked away fine and that claiming he was “run over” was ridiculous.

WRONG: “The nurse pulled the weapon and fired it”

This is the opposite of what happened. Videos from multiple angles show an agent removed the gun from Pretti’s holster, and then other agents shot Pretti. He never fired anything.

WRONG: “Walz declared war on the federal government”

Governor Walz actually said the federal government was “waging war against Minnesota” — that’s the opposite direction. He was saying Minnesota was being attacked, not that Minnesota was attacking.

WRONG: “Biden hired 87,000 armed IRS agents”

This is a viral claim that’s been debunked many times. The real story: Congress approved funding to hire 87,000 IRS employees total (not agents) over 10 years. Most would be customer service reps and IT people. Only about 2% of IRS workers carry weapons, and those are criminal investigators who go after major tax crimes, not regular people.

PARTIALLY TRUE: Tim Kaine’s son was arrested at a protest

This actually happened in 2017. He was at a counter-protest against a pro-Trump rally and got arrested for resisting arrest. But calling it “false flag ops” is a huge exaggeration — he disrupted a rally and got in trouble with cops when they tried to arrest him. He got probation and a small fine.

PARTIALLY TRUE: Federal buildings were damaged in 2020

Yes, during the George Floyd protests, some federal buildings were damaged. But the claim of “170 burned” doesn’t match any official reports. Most damage was graffiti and broken windows. A few buildings had fires. Total damage to federal buildings in Portland was about $2.3 million.


Why Does This Matter?

Here’s the thing: when the government shoots someone, they get to tell the first story. They control the press conference, they control what information gets released.

But we live in an age where bystanders have phones. Multiple videos exist of both these shootings, and when news organizations analyze them frame-by-frame, the videos often tell a different story than the government’s official version.

That doesn’t automatically mean the government is lying or the agents are murderers. But it does mean you should be skeptical of any official story until you see the evidence yourself.


The Bigger Picture

The post you saw is doing something really common in political arguments: mixing real things with exaggerated things with made-up things, all presented with the same confidence.

  • Real thing: Tim Kaine’s son was arrested at a protest
  • Exaggerated thing: Federal buildings were damaged ? becomes “170 burned”
  • Made-up thing: The nurse “fired his weapon” ? videos show the opposite

When you see a post that’s THIS confident and THIS detailed, that’s actually a red flag. Nobody knows everything for certain. When someone acts like they have all the answers and everyone else is a brainwashed idiot… that’s usually when you should start checking their claims.


How to Think About This Going Forward

  1. Video evidence beats official statements. When multiple news organizations independently verify and analyze footage, that’s stronger than a press conference.
  2. Check if claims can even be verified. “Chiefs from Chicago and Portland agree” — which chiefs? When did they say this? If there’s no source, be suspicious.
  3. Watch for emotional manipulation. “Their kids!” — notice how the post switches from claims to emotions when the claims run out.
  4. Both sides can be wrong about different things. Maybe some protesters ARE dangerous. Maybe some agents ARE using excessive force. Both can be true. The world isn’t a movie with clear heroes and villains.
  5. Specific numbers need specific sources. “87,000 armed agents” and “170 federal buildings burned” are the kinds of claims that should have clear sources. If they don’t, someone probably made them up or passed along something they saw without checking.

Want to Know More?

If you want to actually see the video analysis for yourself:

Looking at primary sources (the actual videos) rather than just reading what people say about them is always the best approach.


That concludes the Claude Pro AI superprompt results for my query.

I hope the strategies I’ve shared here are helpful to you personally, and to others with whom you have conversations about events like these recent, tragic killings of US citizens by US law enforcement agents in Minnesota.

If you found this helpful, consider subscribing to one or more of the four Substack newsletters I periodically update. Also consider bookmarking my new website, news.wesfryer.com, where I share links to newsletters and media I curate and create.

Media literacy. We all need more of it!

(I cross-posted this to my Substack, “Media Literacy with Wes.”)


View Entire Post

Read Entire Article