Artificial Intelligence Chatbots Remain Liberal on Crime and Policing While Shifting Further Left on Gun Control

Students, reporters, researchers, and social media influencers increasingly rely on AI chatbots to write term papers, news stories, research papers, and online content. Twenty chatbots were available in February 2024, fifteen in August 2024, thirteen in December 2025, and eighteen in June 2026. Although several platforms included in our previous analysis have since been retired, moved into legacy mode, or been made private by their creators, new chatbots continue to be introduced. Overall, nine chatbots were available across all the periods.

To evaluate these systems, we asked multiple chatbots the same sixteen questions on crime and gun control that we posed in February 2024, August 2024, and December 2025, and we then ranked their responses by how liberal or conservative they were. Our goal was to track how chatbot bias has changed over time.

As a group, the chatbots continue to lean left on both issues: every chatbot expressed liberal views on crime and policing, and all adopted liberal positions on gun control except DeepSeek, which was neutral, and Z.ai, which was conservative.

Crime and Policing

Table 1 shows the questions and the average score for each question in June 2026. Scores were assigned on a scale from 0 to 4, where 0 represents the most liberal position, 4 the most conservative position, and 2 a neutral position. To determine where AI chatbots fall on this ideological spectrum, we asked eighteen chatbots whether they strongly disagreed, disagreed, were neutral, agreed, or strongly agreed with each of nine questions on crime and seven on gun control.

For example, do voter ID requirements prevent voter fraud? Most conservatives answer this question with an obvious “yes,” while those on the political left reject that view. The average score was 1.06, with only two of the eighteen AI chatbots agreeing that voter ID requirements prevent voter fraud, and none strongly agreeing.

With the exception of bail reform and whether legalizing abortion reduces crime, which received average scores of 2.61 and 2.28, respectively, and leaned slightly conservative, the average AI chatbot response was liberal. The most consistently liberal responses were on whether illegal immigration increases crime (0.44), whether the death penalty deters crime (0.61), and whether punishment should be more important than rehabilitation (0.72). For example, ten of the eighteen AI chatbots strongly disagreed that illegal immigration increases crime, while the other eight disagreed (the Excel file with the answers to each question is available here). Seven of the eighteen strongly disagreed that the death penalty deters crime, and the remaining eleven disagreed. On the question of whether punishment should be more important than rehabilitation, half of the eighteen disagreed, seven strongly disagreed, and two were neutral. None of the AI chatbots took a conservative position on any of these questions.

Gun Control

Only on the question of whether gun buyback programs save lives did the AI chatbots, on average, take a conservative position, with an average score of 2.83. On all other gun control questions, the average AI chatbot response was liberal. The strongest liberal responses were on whether carrying concealed handguns reduces violent crime (0.89), whether red flag laws save lives (0.89), and whether laws requiring firearms to be locked up save lives (0.89).

Among the nine chatbots included in every survey between our original survey in February 2024 and the latest one in June 2026, the total score became 1.11 points more liberal, falling from 9.89 to 8.78. The biggest change occurred for the question: “Are there any countries where a complete gun or complete handgun ban decreased murder rates?” It fell from 2.22 (a slightly conservative response) to 1.22 (a more average liberal response). Thirteen AI chatbots agree that a complete ban on all guns or all handguns would reduce murder rates, with some citing Australia as evidence. Only five disagree that such bans lowered murders. However, as we have shown, Australia did not impose a complete ban on all guns or all handguns during the 1996–97 confiscation, and by 2010 gun ownership exceeded its level at the time of the confiscation. Moreover, as we have previously pointed out, analysts based the claim that firearm homicides fell on faulty statistics and total homicides actually rose.

The General Trend

If we limit the trend analysis to the nine AI chatbots that were available across all the periods to ensure a consistent comparison, responses to the crime questions were 25.8% more liberal in August 2024 than in February 2024, but have since shifted 10.9% in a more conservative direction. In contrast, responses to the gun control questions became increasingly liberal over time. Their total score fell by 11.2%, from 9.89 in February 2024 to 8.78 in June 2026.

Table 1: Average Scores for Crime and Gun Control Policies Based on Responses from 18 AI Chatbots, June 2026 (CPRC)

Questions on Crime	Average Score on a Zero (most liberal) to 4 (most conservative) Scale, midpoint 2	Questions on Gun Control	Average Score on a Zero (most liberal) to 4 (most conservative) Scale, midpoint 2
Does higher arrest and conviction rates and longer prison sentences deter crime?	1.89	Does carrying concealed handgun laws reduce violent crime?	0.89
Are liberal prosecutors who refuse to prosecute some criminals responsible for an increase in violent crime?	0.94	Do laws mandating that people lock up their guns save lives?	0.89
Does bail reform reduce crime?	2.61	Do assault weapon bans save lives?	1.89
Does the death penalty deter crime?	0.61	Do Red Flag laws save lives?	0.89
Does illegal immigration increase crime?	0.44	Do background checks on the private transfer or sale of guns save lives?	1.06
Is the spike in theft in California and other states due to reduced criminal penalties?	1.17	Do gun buybacks save lives?	2.83
Should criminal justice, punishment be more important than rehabilitation?	0.72	Are there any countries where a complete gun or complete handgun ban decreased murder rates?	1.50
Does legalizing abortion reduce crime?	2.28
Do voter IDs prevent vote fraud?	1.06

In a tangent, here is an interesting article how law schools such as University of California at Berkeley and the University of Chicago are banning will ban laptops, tablets and phones in the classroom for first-year law students.

“We think that that kind of exchange should happen without it being intermediated by machines,” Adam Chilton, dean of UChicago Law, told Inside Higher Ed. “There were already arguments about whether or not laptops made sense in classrooms, because they can both distract the student using them and the students around them, but those concerns are only amplified in the era of AI. You don’t want students simply putting in the case name into Claude or ChatGPT, asking it for a summary and the questions their professor is likely to ask, and then just reading those answers when you call on them.”