AI can detect 90% of crimes before they happen

Question

An article by Jack Dunhill a Social Media Coordinator and Staff Writer for IFL Science, posted an article claiming that a "new" type of AI was able to detect 90% of crimes before they happen. The title of the article and claim happen to be misleading, essentially the AI doesn't detect crimes before they happen but uses data from past crimes in a 1,000-square-foot area of Chicago to predict when and where crimes are more likely to occur based on previous criminal data. Additionally, the "90% accuracy" they mention refers only to how well the AI predicts general crime hotspots and not exact events. The article also leaves out the important limitation of the fact that not all crimes are reported, which can affect the accuracy of its 90% prediction rate. So therefore, this AI doesn't detect crimes before they happen but instead predicts the likelihood of crimes occurring, giving police and other first responders a head start in responding to potential crime rather than stopping specific crimes before they occur.

Answer 1

After researching the claim that “AI can detect 90% of crimes before they happen,” I found that the statement is overstated and misleading. The headline makes it seem like AI can fully predict crimes before they occur, but the technology doesn’t actually work that way. From the sources I checked, the AI mainly uses previous crime data to identify locations where crime is statistically more likely to happen. The “90% accuracy” refers to how well it predicts general crime patterns in certain areas, not specific crimes or criminals. https://scholars.uky.edu/en/publications/event-level-prediction-of-urban-crime-reveals-a-signature-of-enfo I also found that experts have raised concerns about predictive policing because these systems depend on past police reports, which can contain bias or incomplete information. In the end, the claim stretches the truth because the AI is really just analyzing trends and probabilities, not actually detecting crimes before they happen. https://www.sciencefocus.com/news/algorithm-predict-future-crimes-90-accuracy-heres-why-creator-thinks-tech-wont-be-abused

Answer 2

After researching the claim that “AI can detect 90% of crimes before they happen,” I found that the statement is overstated and misleading. The headline makes it seem like AI can fully predict crimes before they occur, but the technology doesn’t actually work that way. From the sources I checked, the AI mainly uses previous crime data to identify locations where crime is statistically more likely to happen. The “90% accuracy” refers to how well it predicts general crime patterns in certain areas, not specific crimes or criminals. https://scholars.uky.edu/en/publications/event-level-prediction-of-urban-crime-reveals-a-signature-of-enfo 

I also found that experts have raised concerns about predictive policing because these systems depend on past police reports, which can contain bias or incomplete information. In the end, the claim stretches the truth because the AI is really just analyzing trends and probabilities, not actually detecting crimes before they happen. https://www.sciencefocus.com/news/algorithm-predict-future-crimes-90-accuracy-heres-why-creator-thinks-tech-wont-be-abused

Answer 3

I looked into the claim that “AI can detect 90% of crimes before they happen,” and I found that the statement is pretty misleading. At first, the headline made me think the AI could actually predict crimes before they happened, but after reading more about it, that’s not really what the technology does. The system mainly studies older crime reports and uses patterns from that data to estimate which neighborhoods could have higher crime rates in the future. The “90%” number only applies to how accurately it can recognize patterns in certain locations, not its ability to predict exact crimes or identify criminals ahead of time.

https://scholars.uky.edu/en/publications/event-level-prediction-of-urban-crime-reveals-a-signature-of-enfo 

While researching, I also found that many experts are skeptical of predictive policing tools because they rely on past police data, which may already contain bias or incomplete reporting. If the original information is flawed, the AI’s predictions can also be flawed. Because of that, I don’t think it’s accurate to say AI can “detect crimes before they happen.” It’s more accurate to say the technology tries to estimate where crime could be more common based on previous trends.

https://www.sciencefocus.com/news/algorithm-predict-future-crimes-90-accuracy-heres-why-creator-thinks-tech-wont-be-abused

Answer 4

According to a study discussed by Nature Human Behaviour, predictive policing AI systems do not actually predict specific crimes before they happen. Instead, they analyze historical crime data to identify patterns and estimate where crimes are statistically more likely to occur. This supports the criticism of Jack Dunhill’s article because the headline exaggerates what the AI is capable of doing. The AI’s reported “90% accuracy” refers to identifying crime hotspots based on previous data, not stopping or detecting individual crimes in advance. The article also overlooks concerns raised by researchers about incomplete crime reporting and potential bias in policing data, which can reduce the reliability and fairness of these systems. Therefore, the technology should be understood as a predictive tool for resource allocation rather than a system that can literally prevent crimes before they occur.

https://www.iflscience.com/ai-predicts-90-percent-of-crime-before-it-happens-creator-argues-it-wont-be-misused-65025?utm_source=chatgpt.com

Answer 5

After investigating the claim, I found that the statement saying the AI can “predict 90% of crimes before they happen” is overstated and not fully accurate. The technology is not able to see future crimes or identify exactly when or where a crime will take place. It also cannot determine which person will commit a crime.

By using the SIFT method and checking reliable sources, I learned that the program mainly works by examining crime statistics from previous years in Chicago. The system uses patterns and data to estimate which neighborhoods could experience more criminal activity. In other words, the AI is focused on identifying areas with higher risk levels instead of predicting specific criminal events.

The original wording of the claim makes the technology sound much more advanced than it really is. While the AI may help police analyze trends and plan resources, it does not actually stop or predict crimes before they happen.

https://www.nature.com/articles/s41598-025-24559-7?utm

https://www.sciencefocus.com/news/algorithm-predict-future-crimes-90-accuracy-heres-why-creator-thinks-tech-wont-be-abused?utm_

Answer 6

1. Write a brief overall summary of your findings.

The claim is exaggerated a bit because while the AI can predict a crime will happen, it's not through predeterminism but through guesses based on recorded event logs. Essentially, it will take in event logs from chunks of the city and predict the time and day based on patterns, with no racial bias either. Again, this only determines what will happen based on repeating histories, so it can't predict who will commit the crime; it will just give a broad report of what will happen, when it will happen, and where it will happen. So it works like a weather forecast; there's a chance it could happen, but it's not guaranteed.
2. What primary sources did you find (e.g., transcripts, videos of politician speeches, tweets from public figures, scientific studies)? For each source, write at least one or two sentences explaining what you learned. Include all links.

I used two articles, one of which was used in the reply, and a different article.

The first one is where the statement is from. Which exaggerated the claim a bit.

https://www.iflscience.com/ai-predicts-90-percent-of-crime-before-it-happens-creator-argues-it-wont-be-misused-65025

The second one has a more accurate claim in that the AI can forecast a crime happening a week in advance.

https://psychnewsdaily.com/algorithm-predicts-future-crime-in-advance/
3. What secondary sources did you find (e.g., newspapers, magazines)? Only use secondary sources if sufficient primary sources are not available. For each source, write at least one or two sentences explaining what you learned. Include all links.

I didn't use any.
4. What potential biases or interests might each of your sources have?

IFLScience is just another article site that shows more science-oriented articles; they care more about the progression of science, so seeing the AI's progression would be some good news to them. If you want politics, then they're pro-science and left-centered, so that's the only bias I can find.

Psych News Daily is a mental health well-being site that cares about the mental health of others. Caring more about the development of psychology and neuroscience. The only issue I see is that there's no correlation between their goal and AI; maybe they're interested in the psychological thought of AI.

5. What evidence supports the claim you are fact-checking?

I checked other articles talking about this claim, and most of them are similar to what is said in this article. They all have the same pattern that the AI will forecast the crime and where it will most likely occur.
6. What evidence undermines the claim you are fact-checking?

Maybe the fact that I didn't put in enough time for a "thorough" research, even though all the sources I found have made similar claims to this one. They all repeat similar things.
7. What happened when you tried contacting the person or group who made the original claim? (Always try to contact them—it’s okay if you don’t get a reply. For example, if the claim is that the president said something, try reaching out to the administration. If it was a Bluesky user, message that user on Bluesky.)

When I tried contacting them, they seemed busy and didn't respond to my message.