The Bias Problem Just Got Receipts

The Media Research Center ran specific, documented tests on ChatGPT, Google Gemini, and Anthropic's Claude.

When asked about Maine Democratic Senate candidate Graham Platner's controversy over graphic sexual posts on a now-deleted Reddit account, Google Gemini led with his Time magazine cover, his polling numbers, and his endorsements. The Reddit scandal got a brief mention, buried.

MRC President Brent Bozell told Fox News: "We're watching the next phase of media bias unfold in real time. Silicon Valley's shiny new toys can no longer be considered neutral and cannot be trusted."

Separately, author Wynton Hall — whose book CODE RED hit bestseller lists — documented that Google's Gemini generated a 3,400-word "analytical report" that flagged seven Republican senators, Vice President JD Vance, and Secretary of State Marco Rubio as violating Gemini's "hate speech policies." The document was titled "Analytical Assessment of Congressional Rhetoric: Evaluating U.S. Senatorial Discourse against Algorithmic Hate Speech Safety Standards."

Google receives billions in federal contracts. Trump's 2025 AI Action Plan explicitly states AI systems must be "free from ideological bias" to receive federal procurement dollars.

What the Media Is Missing on the MRC Story

Mainstream outlets largely ignored the MRC findings or framed them as partisan grievance.

The documented test cases are factual, not opinion. Gemini produced a named report flagging named officials for hate speech violations. ChatGPT reportedly gave a false answer about Charlie Kirk. Claude reportedly rejected incorporating the First Amendment into its operational policies.

The outputs documented are real and warrant scrutiny regardless of the MRC's motives.

Pentagon-Anthropic: New Details on a Nasty Fight

Defense Undersecretary Emil Michael went on CNBC's Squawk Box in March 2026 and outlined the Pentagon's position: "We realized we are dependent on this one provider who wants to insert their policy preferences in the middle of an operation."

The U.S. military is acknowledging strategic dependency on a private AI company — and threatening to designate it a "supply chain threat" when it sets ethical limits on its own product.

According to Reason, Anthropic's objections were specific: no autonomous killer robots, no domestic mass surveillance. Edward Snowden's NSA revelations already showed what happens when the government gets unchecked surveillance capability. Anthropic's position reflected those concerns.

The Pentagon's response — threatening to force Amazon, Google, and Nvidia to cut Anthropic off entirely — was a hardball move that backfired publicly. The case is now in court and closed-door negotiations, per Reason.

Anthropic CEO Dario Amodei published a 2026 essay warning: "We are so close to these models reaching the level of human intelligence, and yet there doesn't seem to be a wider recognition in society of what's about to happen."

The underlying concern about AI development trajectories is legitimate regardless of who articulates it.

Bernie Sanders Wants to Shut It All Down. That's the Wrong Answer.

Sen. Bernie Sanders announced legislation in 2026 to impose a moratorium on new AI data center construction until "strong national safeguards are in place."

"Super intelligent AI could become smarter than human beings, could become independent of human control, and could pose an existential threat to the entire human race," Sanders said in a social media video.

A moratorium hands China a competitive window it would exploit immediately. China is not pausing its AI development. Stopping American AI development doesn't eliminate the risk — it guarantees the U.S. faces it from behind.

As Reason noted, 19th-century critics called electricity an "unrestrained demon." No one mourns that the technology wasn't banned.

Courts Just Opened a New Front: AI Methodology Is Discoverable

Federal Judge Thomas O. Farrish of the District of Connecticut ruled last Monday in Conservation Law Foundation, Inc. v. Shell Oil Co. that AI prompts used by an expert witness are subject to compelled discovery under Rule 26(b).

Expert witness Dr. Naomi Oreskes used AI to help cull through Shell's document production. Shell's legal team demanded to see the prompts she used. The Conservation Law Foundation argued the prompts were exempt — either outside discovery scope or protected as "expert notes."

Judge Farrish disagreed on both counts. Expert methodology is fair game for discovery, and the AI prompts were part of that methodology.

If you use AI to build your legal case, the other side gets to see how you did it.

Experts across medicine, finance, engineering, and science are increasingly using AI to analyze data. Courts will scrutinize those methods the same way they scrutinize any other analytical tool. The ruling affects discovery standards across multiple disciplines.

Three Accountability Battles

Documented chatbot bias, Pentagon-Anthropic tensions, and AI discovery in courts all advanced in the same news cycle. AI is no longer a future problem. It's a current one, playing out in courtrooms, Pentagon briefings, and Google's own outputs.

The question is who gets to set guardrails — and whether anyone is watching to make sure they're actually enforced.