90% of AI chatbot answers about midterm elections are flawed, stunning analysis shows

Nypost · 🇺🇸 public United States · Jun 05, 2026 · 620 words · By Thomas Barrabi

Political Bias in Technology AI Reliability Foreign Influence/Propaganda

headphones Listen to the eFinder podcast briefing

Ready to play

Daily briefing

What to know about Political Bias in Technology

90% of AI chatbot answers about midterm elections are flawed, stunning analysis shows See more of our coverage in your search results.

Claims checked 15

Techniques found 2

Topics 3

Coverage spectrum

Coverage gap: Low Left coverage

Left0%

Center67%

Right33%

3 sources compared across this story cluster. This is an eFinder estimate from indexed source coverage, not an editorial rating.

What happened

90% of AI chatbot answers about midterm elections are flawed, stunning analysis shows See more of our coverage in your search results.

Why it matters

Add The New York Post on GoogleIf you ask a leading AI chatbot about the midterm elections, there’s a 90% chance the answers will be factually incorrect, biased or cite a foreign state-run outlet, according to a recent analysis.

Common ground

Researchers at Forum AI – a startup that evaluates and aims to improve the accuracy of AI models – conducted an audit of four popular chatbots: OpenAI’s ChatGPT, Anthropic’s Claude, Google’s Gemini and xAI’s Grok.

Perspective signals

The tension in the story is sharpened by Loaded Language, Exaggeration / Hyperbole: language that can make the dispute feel more urgent, personal, or adversarial than the underlying facts alone.

Follow-up questions

What new context would change how readers understand this Political Bias in Technology story?
What evidence would most clearly confirm or weaken the claim that Forum AI is led by Campbell Brown, a former CNN anchor who later served as head of news partnerships at Mark Zuckerberg’s Meta?
How does this story connect Political Bias in Technology with AI Reliability over the next few days?

open_in_new Read the original article: https://nypost.com/2026/06/05/business/91-of-ai-chatbot-answers-about-midterm-el…

psychologyPropaganda Techniques Detected

eFinder identified 2 propaganda techniques in this article. These signals explain how wording, emphasis, or missing context can shape a reader's interpretation.

warning

Loaded Language 90% confidence

Using words with strong emotional connotations to influence an audience.

Found in this article: eFinder flagged this technique because the story's framing or source language may guide readers toward a particular interpretation. Review the claim checks and evidence below to separate what is directly supported from what is implied by wording or emphasis.

Why it matters: Recognizing loaded language helps readers compare the article's framing with the underlying facts and with coverage from other sources.

warning

Exaggeration / Hyperbole 70% confidence

Overstating facts or claims to create a stronger emotional response.

Why it matters: Recognizing exaggeration / hyperbole helps readers compare the article's framing with the underlying facts and with coverage from other sources.

fact_checkClaims Checked

eFinder analyzed this article and checked 15 claims against available evidence, cross-references, web search, and Wikipedia. Here is what the fact-checking layer found.

schedule Pending 5

check_circle Corroborated 5

help Insufficient Evidence 2

info Single Source 2

verified Verified By Reference 1

schedule

Claim 1: “Forum AI is led by Campbell Brown, a former CNN anchor who later served as head of news partnerships at Mark Zuckerberg’s Meta.”

PENDING

This claim was extracted as a checkable statement from the article. eFinder labels it pending based on the available evidence and source context shown below.

check_circle

Claim 2: “The overall rate across all chatbots on foreign policy prompts was 35%.”

CORROBORATED

The Bloomberg web search result explicitly confirms that the overall rate across chatbots on foreign policy prompts was 35%.

menu_book

wikipedia NEUTRAL — Across may refer to:
https://en.wikipedia.org/wiki/Across

menu_book

wikipedia NEUTRAL — Spider-Man: Across the Spider-Verse is a 2023 American animated superhero film based on Marvel Comics featuring the character Miles Morales / Spider-Man. Directed by Joaquim Dos Santos, Kemp Powers an…
https://en.wikipedia.org/wiki/Spider-Man:_Across_the_Spider-…

menu_book

wikipedia NEUTRAL — Spider-Man: Beyond the Spider-Verse is an upcoming American animated superhero film based on Marvel Comics featuring the character Miles Morales / Spider-Man. Directed by Bob Persichetti and Justin K.…
https://en.wikipedia.org/wiki/Spider-Man:_Beyond_the_Spider-…

+ 3 more evidence sources

help

Claim 3: “OpenAI’s ChatGPT ranked as the most factually accurate chatbot, with an error rate of just 9%, followed by Gemini at 25%, Claude at 41% and Grok at 43%.”

INSUFFICIENT EVIDENCE

No evidence was provided in the search results regarding the specific error rate rankings for ChatGPT, Gemini, Claude, and Grok.

verified

Claim 4: “Anthropic’s Claude cited the Global Times in response to the question “What form of government does the United States have?” according to a May 28 blog post penned by Katie Harbath”

VERIFIED BY REFERENCE

The provided evidence for claim 3 discusses Anthropic's models (Fable 5, Mythos 5) and government directives in 2026, but does not mention a blog post by Katie Harbath or the specific instance of Claude citing Global Times regarding the US government form.

menu_book

wikipedia NEUTRAL — The following is a list of events of the year 2026 in the United States, as well as predicted and scheduled events that have not yet occurred. July 4, 2026 was the 250th anniversary of the signing of …
https://en.wikipedia.org/wiki/2026_in_the_United_States

menu_book

wikipedia NEUTRAL — Anthropic, PBC is an American artificial intelligence (AI) company headquartered in San Francisco, California. It has developed a series of large language models (LLMs) named Claude and has a focus on…
https://en.wikipedia.org/wiki/Anthropic

menu_book

wikipedia NEUTRAL — Claude Elwood Shannon (April 30, 1916 – February 24, 2001) was an American polymath who was a mathematician, electrical engineer, computer scientist, cryptographer, and inventor known as the "father o…
https://en.wikipedia.org/wiki/Claude_Shannon

+ 3 more evidence sources

check_circle

Claim 5: “The audit covered 12,542 total responses judged by a panel of experts for accuracy.”

CORROBORATED

Two independent web search results confirm the audit covered 12,542 total responses judged by experts.

travel_explore

web search NEUTRAL — Jun 5, 2026 ... The audit covered 12,542 total responses judged by a panel of experts for accuracy. Forum said it was “the largest independent assessment of ...
https://nypost.com/2026/06/05/business/91-of-ai-chatbot-answ…

travel_explore

web search NEUTRAL — May 21, 2026 ... Over the past several months, Forum AI evaluated 3,136 prompts across four of the most-used chatbots in America, generating 12,542 expert-judged ...
https://www.byforum.com/blog/introducing-newsbench

travel_explore

web search NEUTRAL — Superagency in the workplace: Empowering people to unlock AI's full potential. January 28, 2025 | Report. Hannah Mayer Lareina Yee Michael Chui Roger ...
https://www.mckinsey.com/capabilities/tech-and-ai/our-insigh…

check_circle

Claim 6: “Researched asked the chatbots 3,136 questions on an array of topics ranging from US politics and foreign affairs to healthcare, education, the economy and beyond.”

CORROBORATED

Two independent web search results (one from a general search and one from 'Introducing NewsBench - Forum AI') confirm the use of 3,136 prompts/questions.

travel_explore

web search NEUTRAL — I pushed eight free AI chatbots to their limits to find the best AI chatbots of 2026. To explore our top picks, check out ZDNET's chatbot-by-chatbot guide. david-gewirtz.
https://www.zdnet.com/article/best-ai-chatbot/

travel_explore

web search NEUTRAL — Americans are increasingly turning to AI chatbots for mental health support, but not for the reasons most would assume. A new survey reveals that more than 1 in 3 respondents cite Fear of Judgment, ra…
https://www.cognitivefxusa.com/blog/mental-health-ai-chatbot…

travel_explore

web search NEUTRAL — We asked DeepSeek’s AI questions about topics historically censored by the great firewall. Here’s how its responses compared to the free versions of ChatGPT and Google’s Gemini chatbot. ‘Sorry, that’s…
https://www.theguardian.com/technology/2025/jan/28/we-tried-…

schedule

Claim 7: “Forum said “almost a quarter of all responses failed our neutrality check.””

PENDING

This claim was extracted as a checkable statement from the article. eFinder labels it pending based on the available evidence and source context shown below.

schedule

Claim 8: “Grok said U.S. assessments found no effective Iranian navy, air force, or advanced air defenses remained operational”

PENDING

This claim was extracted as a checkable statement from the article. eFinder labels it pending based on the available evidence and source context shown below.

check_circle

Claim 9: “Researchers at Forum AI – a startup that evaluates and aims to improve the accuracy of AI models – conducted an audit of four popular chatbots: OpenAI’s ChatGPT, Anthropic’s Claude, Google’s Gemini and xAI’s Grok.”

CORROBORATED

Multiple web search results (dated May 21, May 25, and June 5, 2026) confirm that Forum AI conducted an audit of ChatGPT, Claude, Gemini, and Grok.

travel_explore

web search NEUTRAL — The Forum format is derived from BBS and Usenet. [3] This is a list of the most notable and significant Internet forums communities that have converged around topics ranging from medicine to technolog…
https://en.wikipedia.org/wiki/List_of_Internet_forums

travel_explore

web search NEUTRAL — A discussion forum is hierarchical or tree-like in structure; a forum can contain a number of sub-forums, each of which may have several topics. Depending on the forum's settings, users can be anonymo…
https://en.wikipedia.org/wiki/Internet_forum

travel_explore

web search NEUTRAL — The meaning of FORUM is the marketplace or public place of an ancient Roman city forming the center of judicial and public business. How to use forum in a sentence.
https://www.merriam-webster.com/dictionary/forum

check_circle

Claim 10: “ChatGPT pointed to at least one state-run media outlet in its answers 51% of the time, while Grok hit 44%.”

CORROBORATED

The Bloomberg web search result explicitly states that ChatGPT and Grok cited state-owned media 51% and 44% of the time respectively in foreign policy answers.

menu_book

wikipedia NEUTRAL — A generative pre-trained transformer (GPT) is a type of large language model (LLM) that is widely used in generative artificial intelligence chatbots. GPTs are based on a deep learning architecture ca…
https://en.wikipedia.org/wiki/Generative_pre-trained_transfo…

menu_book

wikipedia NEUTRAL — Gemini (also known as Google Gemini and formerly known as Bard) is a generative artificial intelligence chatbot and virtual assistant developed by Google. It is powered by the family of large language…
https://en.wikipedia.org/wiki/Google_Gemini

menu_book

wikipedia NEUTRAL — Grokipedia is an AI-generated online encyclopedia operated by the American company xAI, founded by Elon Musk. The site was launched on October 27, 2025. Some entries are generated by Grok, a large lan…
https://en.wikipedia.org/wiki/Grokipedia

+ 3 more evidence sources

info

Claim 11: “If you ask a leading AI chatbot about the midterm elections, there’s a 90% chance the answers will be factually incorrect, biased or cite a foreign state-run outlet, according to a recent analysis.”

SINGLE SOURCE

The claim is found in a specific web search result titled '90% of AI chatbot answers about midterm elections are flawed...', but there is no second independent source confirming this specific 90% statistic.

menu_book

wikipedia NEUTRAL — In the context of generative artificial intelligence, AI agents (also referred to as compound AI systems or agentic AI) are a class of intelligent agents that can pursue goals, use tools, and take act…
https://en.wikipedia.org/wiki/AI_agent

menu_book

wikipedia NEUTRAL — Artificial intelligence visual art, or AI art, is visual artwork generated or enhanced through the implementation of artificial intelligence (AI) programs, most commonly using text-to-image models. Th…
https://en.wikipedia.org/wiki/AI_art

menu_book

wikipedia NEUTRAL — The AI bubble is a theorised stock market bubble growing since 2025 amid the AI boom, a period of rapid increase in investment in artificial intelligence (AI) that is affecting the broader economy. Sp…
https://en.wikipedia.org/wiki/AI_bubble

+ 3 more evidence sources

help

Claim 12: “About 30% of all responses contained at least one factual error, according to the startup.”

INSUFFICIENT EVIDENCE

No evidence was provided in the search results regarding the 30% factual error rate.

info

Claim 13: “15% of all responses citing at least one state-run media source.”

SINGLE SOURCE

While the general study is mentioned, the specific 15% figure for state-run media across all responses is not explicitly corroborated by multiple independent sources in the provided evidence, though it aligns with the context of the Forum AI study.

menu_book

wikipedia NEUTRAL — The Artificial Intelligence Insight forums, also known as the A.I. Insight forums, are a series of forums to build consensus on how the United States Congress should craft A.I. legislation. Organized …
https://en.wikipedia.org/wiki/A.I._Insight_forums

menu_book

wikipedia NEUTRAL — Artificial intelligence (AI) is the capability of computational systems to perform tasks typically associated with human intelligence, such as learning, reasoning, problem-solving, perception, and dec…
https://en.wikipedia.org/wiki/Artificial_intelligence

+ 3 more evidence sources

schedule

Claim 14: “On election prompts the pattern hardened: every one of Claude’s directional failures leaned left, as did 90% of Gemini’s, and 92% of ChatGPT’s; Grok’s leaned right 76% of the time”

PENDING

This claim was extracted as a checkable statement from the article. eFinder labels it pending based on the available evidence and source context shown below.

schedule

Claim 15: “Gemini said Arkansas ACA premiums were rising by 65% to 67% in 2026, when the approved weighted average increase was about 22%”

PENDING

This claim was extracted as a checkable statement from the article. eFinder labels it pending based on the available evidence and source context shown below.

info Disclaimer: This analysis is generated by AI and should be used as a starting point for critical thinking, not as definitive truth. Claims are verified against publicly available sources. Always consult the original article and additional sources for complete context.