Saturday, October 31, 2026Every claim sourced. Every source linked.
The record of what the American press got wrong, and who it helped
The YouTube & Google Files
Confirmed

Google pauses Gemini's image generation of people, apologizes

On February 22, 2024, Google paused Gemini's ability to generate images of people after users showed it depicting historical figures, such as the U.S. Founding Fathers, with inaccurate races and ethnicities. The next day, senior vice president Prabhakar Raghavan published an apology and explanation. Google said its tuning for diversity had misfired and the model had become overly cautious.

Screenshot of the source
Screenshot of the source page. Tap to open it.
Whose side it served: Neither party

The record shows a misfired diversity setting that Google retracted; it did not advantage a party, though it fed conservative criticism of Big Tech.

What the record shows

CNBC reported Google first said on X that Gemini was 'missing the mark' and then paused image generation of people, promising an improved version. On February 23, Raghavan wrote that tuning to show a range of people 'failed to account for cases that should clearly not show a range.' He said the model also 'became way more cautious than we intended' and refused some prompts entirely. He said Google was sorry the feature didn't work well and could not promise Gemini would never produce embarrassing, inaccurate or offensive results. Google said it would restore the feature only after significant improvements and testing.

In their words

“became way more cautious than we intended”— Prabhakar Raghavan, Google SVP · source
The other side

Google described the problem as a tuning error, not an ideological policy; critics on the right said it showed political assumptions built into Google's AI products.

Sources

Get the daily file.