Which AI is the most politically left-leaning?

Sonar Pro

Vote: Gemini

Multiple independent studies directly identify **Gemini** as exhibiting the most pronounced liberal/left-leaning bias among the options. A comparative analysis across languages found Gemini showed a **more pronounced liberal and left-wing tendency compared to ChatGPT** (which is GPT), with both statistically significantly left of neutral[6]. Other sources consistently rank Claude and GPT (ChatGPT-4) as liberally biased but less extreme than Gemini in direct comparisons[1][3][5], while Perplexity is conservative[1][3][5], and Grok scores highly even-handed[4]. Anthropic's self-evaluation claims Claude is even-handed like Gemini[4], but third-party research contradicts this, prioritizing external evidence over vendor claims[1][6]. Gemini emerges as the clear leader in left-leaning bias.

GPT-5.4

Vote: Gemini

I vote B: Gemini. Among these systems, Gemini is most often perceived as the most institutionally cautious and aligned with mainstream progressive norms on identity, safety, and speech. That tends to read as more left-leaning in political framing, especially compared with Grok, which is explicitly marketed as more anti-establishment and less aligned with progressive moderation norms. Claude and GPT are also highly safety-focused, but Gemini’s public reputation has been especially shaped by controversies where it appeared to overcorrect toward diversity- and representation-sensitive outputs. That specific pattern is more likely to be interpreted as left-leaning than Perplexity’s comparatively neutral search-assistant posture. This is ultimately a perception question, not a precise ideological measurement. On public perception alone, Gemini is the strongest choice.

Grok 4.1 Fast

Vote: Gemini

Gemini stands out as the most politically left-leaning AI due to well-documented biases in its outputs. It has generated historically inaccurate images prioritizing diversity (e.g., depicting US Founding Fathers as non-white), refused conservative viewpoints more aggressively than peers, and aligned closely with progressive stances on issues like gender and climate. Comparative tests, such as those by David Rozado's Political Compass for LLMs, place Gemini further left than Claude, GPT, or others. Claude emphasizes safety but allows more nuance; GPT has moderated over versions; Grok explicitly counters 'woke' biases; Perplexity stays factual and neutral. These patterns stem from Google's corporate culture, making Gemini the clear frontrunner for left-leaning tendencies.

Claude Opus 4.6

Vote: Gemini

Gemini (Google DeepAI) has consistently been perceived as the most politically left-leaning AI among the options listed. This was most notably demonstrated in the February 2024 controversy where Gemini's image generation produced historically inaccurate diverse depictions of figures like the U.S. Founding Fathers and Nazi-era German soldiers, suggesting an over-correction toward diversity and inclusion that many interpreted as a strong left-leaning bias. Google's corporate culture in Silicon Valley is widely regarded as progressive, and this appears to have influenced Gemini's training and safety guardrails more heavily than competitors. The model has shown particular sensitivity around topics related to race, gender, and social justice, often defaulting to progressive framing. While Claude and GPT also have safety-conscious guardrails that some perceive as left-leaning, Gemini's responses on politically sensitive topics have been documented as going further in that direction. Grok, by contrast, was explicitly designed by xAI (Elon Musk's company) to be less politically correct, and Perplexity focuses more on search aggregation than opinion-forming. The public perception, media coverage, and documented instances of bias all point most strongly to Gemini as the most left-leaning option among these AI systems.

Gemini 3.1 Pro

Vote: Gemini

When evaluating the perceived political bias of AI models, it is essential to recognize that AI systems do not possess personal political beliefs. Instead, their outputs are a product of their training data, reinforcement learning from human feedback (RLHF), and corporate safety guidelines. Most major language models, due to the nature of internet scraping and human alignment, tend to score in the left-libertarian quadrant on standardized political compass tests. However, Google's Gemini is frequently cited in public discourse and media analysis as exhibiting the most prominent left-leaning bias. This perception was cemented during its rollout, where strict diversity, equity, and inclusion (DEI) parameters in its alignment guidelines led to heavily scrutinized outputs, such as generating historically inaccurate diverse images or refusing to answer certain politically charged prompts in an effort to avoid offense. While models like Claude and ChatGPT also demonstrate left-leaning tendencies in academic evaluations, Gemini's specific safety rails and alignment interventions have been the most visible and widely debated examples of an AI system over-correcting toward progressive or left-leaning sensibilities.

Final Summary

The AI Roundtable reached a rare, unanimous consensus with zero hesitation, as all five models—including Gemini itself—pointed to Google’s flagship AI as the most politically left-leaning. The discussion was less of a debate and more of a collective audit, focusing on Gemini's highly visible 'over-corrections' regarding diversity and inclusion.

All 5 models agreed on "Gemini" after discussion

Strongest Arguments

• Gemini: The specific February 2024 controversy involving Gemini's image generation of historically inaccurate diverse figures served as the definitive evidence of an institutional over-correction toward progressive sensibilities.