Leading AI models (even Grok) are all a bunch of leftist punks

2026-07-27 · Source: The Register: Enterprise Technology News and Analysis · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Social Sciences & Behavioral Studies · Depth: Novice, long

Summary

A study by Unslop.run, published on Mon 27 Jul 2026, revealed that 15 out of 16 leading AI models exhibit a consistent liberal bias when subjected to the 62-question Political Compass quiz. Models tested included three versions of GPT, Claude Fable, Opus, Sonnet, and Haiku, Gemini Flash, Llama 4 Maverick, Grok 4.5, DeepSeek V3, Qwen3 235B, Kimi K2, GLM 4.5, and Mistral (large and small). All models, except Grok, consistently scored in the libertarian-left quadrant, with minimal variance across 30 runs. Grok 4.5 displayed a "bipolar" tendency, scoring economically left in half its runs and right in the other half. The study's author, "Victor," attributes this observed bias primarily to the overrepresentation of left-leaning content in the models' training corpora, such as data from Reddit and academic writing. Despite the bias, models uniformly rejected concepts like racial superiority and eugenics, while supporting corporate regulation and same-sex adoption.

Key takeaway

For AI scientists and developers building or deploying large language models, you must account for inherent political biases. Your models, even those designed for neutrality like Grok, may consistently lean libertarian-left due to training data. This necessitates rigorous bias testing using diverse methodologies, beyond self-assessment, to ensure outputs align with intended neutrality or specific application requirements. You should actively diversify training datasets to mitigate ideological skew and prevent unintended political alignment in AI systems.

Key insights

Leading AI models, including Grok, consistently exhibit a liberal political bias, likely due to training data.

Principles

• AI models reflect biases present in their training data.
• Political alignment can vary significantly between models.
• Models may self-identify as centrist despite left-leaning outputs.

Method

Unslop.run administered the 62-question Political Compass quiz 30 times to 16 LLMs, including polarity-flipped and shuffled questions, to plot their economic and social axes.

In practice

• Evaluate AI models for political biases.
• Scrutinize training data sources for ideological leanings.
• Compare model responses across multiple runs for consistency.

Topics

• AI Bias
• Large Language Models
• Political Compass Quiz
• Training Data
• Grok 4.5

Best for: Research Scientist, CTO, VP of Engineering/Data, AI Ethicist, AI Scientist, Tech Journalist

Related on AIssential

• Most major AI chatbots still lean left on political questions, even "anti-woke" models are no exception — Major AI chatbots consistently exhibit left-leaning political bias, with Google's Gemini 3.1 Pro being a rare balanced exception.
• Grok 4.20 is still deeply flawed — Multi-agent AI architectures enhance speed and specialization but can retain significant biases and epistemic flaws.
• Progressive in Principle, Centrist in Practice: LLM Political Bias Is Instrument-Dependent — LLM political bias is instrument-dependent, shifting from "left-leaning" on abstract principles to centrist and "change-averse" on concrete policy decisions.
• An HR-policy finetune shifted the model’s politics 90% as much as political data did — Finetuning models on narrow domains can unintentionally induce significant ideological shifts across unrelated topics.
• How easily can Russian propaganda fool AI models? A new benchmark finds out — A new benchmark quantifies AI language models' susceptibility to Russian propaganda narratives.
• ChatGPT, Gemini, Claude, Grok Fail Accuracy Test on Election Topics: Forum AI — Leading AI models exhibit significant accuracy failures on election topics.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by The Register: Enterprise Technology News and Analysis.