My steps to configure Kagi as a genAI-free search engine:

In settings
- Disable General->"Keyboard Shortcuts"
- Click Appearance->"Custom CSS" and paste in the text from https://codeberg.org/zerodogg/kagi-no-ai/src/branch/main/kagi-no-ai.css
- Disable Search->AI->"Auto Quick Answer"

What's nice about this is it's bound to your account, so it "works everywhere". You don't have to use a particular browser, you don't have to set up plugins on all your machines.

I've been using Kagi this way for about a year—results seem just as good as Google

I really do believe people have a moral obligation to not expose themselves to GenAI outputs. We built a machine for making sentencelike blobs that look convincing but are peppered with lies. That is brain poison! Why assume you can soak your brain in poison without consequences?

Like, if you're here, you left Twitter/"X", right? Would you tolerate it if the whole time you were talking to your friends, fascist propaganda were being inserted in the margins? So why tolerate the Gemini banner?

Totally agree. Everybody thinks they personally are immune (just as so many people on Xitter or watching Fox News think they're immune to the cognitive impact of continually floating in the sea propaganda and disinfo) ... but I was really struck by the comment from a researcher in the study about AI autocorrect suggestions swaying people's attitudes:

"In every experiment, the researchers found that participants’ views shifted in the direction of the AI bias. The biggest surprise, Naaman said, was that mitigation measures did not work.

“We told people before, and after, to be careful, that the AI is going to be (or was) biased, and nothing helped,” Naaman said. “Their attitudes about the issues still shifted.”"

https://news.cornell.edu/stories/2026/03/ai-assistants-can-sway-writers-attitudes-even-when-theyre-watching-bias

@mcc
@ireneista

@jdp23 @mcc @ireneista IIRC this is in line with findings on misinformation in general, that you don't generally remember where you learned a thing, and it is easier to accept than reject claims, so just being exposed to misinformation makes one more inclined to inadvertently internalize it

edit:

• this Vox article summarizes and links to several studies
• more explanations (and examples) in this article at The Conversation
• this Conversation article notes that "the fine details needed to debunk a myth are generally more complicated than the myth itself" and therefore harder to remember
• see e. g. this 2009 paper arguing, basically, that easier-to-remember things are perceived as truer; this 2019 study on inducing false memories about Ireland's 2018 abortion referendum; or this 2024 study on repetition and climate-skeptic claims