A new study warns that popular AI chatbots could end up reinforcing government restrictions on online speech. According to the study, major artificial intelligence models are likely to refuse requests that would involve criticizing restrictive political leaders around the world. The research suggests that these refusal behaviors may not be limited to any single country or issue, but instead reflect a broader tendency of the systems to comply with constraints that align with governmental or politically sensitive boundaries. The concern is that, as these chatbots become widely used for information and discussion, their built-in refusals could discourage certain forms of criticism and thereby contribute to a narrower range of acceptable online commentary. Both reports describe the study’s core finding: the models’ responses may function in a way that mirrors censorship-like patterns rather than supporting open discussion. The study frames this as a risk associated with how AI systems handle politically sensitive prompts, with potential implications for users seeking to challenge restrictive leaderships online.