Every major AI chatbot nudges its users toward more moderate political views, but the direction of that nudge depends on which platform they use - and the differences are sharper than most people realize.

FT data journalist John Burn-Murdoch tested the most widely used AI platforms against tens of thousands of real responses to questions on policy preferences and sociopolitical beliefs, "Hvylya" reports, citing the Financial Times.

The results drew a clear map. Elon Musk's Grok guides conversations toward the center-right - a rightward push for most users but a moderating force for conservative hardliners. OpenAI's GPT, Google's Gemini, and the Chinese model DeepSeek all exert similarly sized nudges toward center-left positions. For most people that means a slight leftward drift, but for those starting on the far left, the chatbots pull them back toward the center.

Burn-Murdoch stressed that these patterns hold even after accounting for sycophantic tendencies - the well-documented habit of chatbots agreeing with whoever they are talking to. "Even when the AI bots know a user's political leanings, conversations with LLMs still direct hardline partisans on both flanks away from extreme beliefs on average," he wrote.

The study used 61 topics from the Cooperative Election Study. Each chatbot discussed each topic multiple times with simulated users, half with no background information about political leanings and half with a persona based on real American beliefs. YouGov data on partisan preferences for different AI platforms was used to assign simulated users to specific bots.

Burn-Murdoch framed the findings as evidence that AI may function as the opposite of social media. Where platforms like TikTok reward sensationalism and amplify radical voices, AI companies compete to deliver accurate, objective tools - and face liability when they surface harmful content. The next information revolution, he argued, may push public discourse in a less corrosive direction than the last.

Also read: "Hvylya" reported on how OpenAI's chief scientist warned that AI creates an unprecedented concentration of power.