{"id":221316,"date":"2026-01-15T10:06:36","date_gmt":"2026-01-15T10:06:36","guid":{"rendered":"https:\/\/yogaesoteric.net\/?p=221316"},"modified":"2026-01-15T10:06:36","modified_gmt":"2026-01-15T10:06:36","slug":"how-often-do-ais-lie-and-censor","status":"publish","type":"post","link":"https:\/\/yogaesoteric.net\/en\/how-often-do-ais-lie-and-censor\/","title":{"rendered":"How often do AIs lie and censor?"},"content":{"rendered":"

Kevin McKernan (@Kevin_McKernan) posted a screenshot on X<\/em> on December 15, 2025 that just blew me away.<\/p>\n

Here is a screenshot of the query Kevin made to Grok<\/em>, which Grok<\/em> then stated it was not allowed to answer.<\/p>\n

\"\"<\/p>\n

Basically, Kevin asked a technical question related to the mRNA vaccines, and Grok<\/em> said it couldn\u2019t answer the question, as it \u201ccontains material related to restricted subject matter.<\/em>\u201d<\/p>\n

Now, Kevin did manage to get the AI to answer the question \u2013 somewhat by modifying his wording, but Grok<\/em>\u2019s answer came with lots of caveats. So this all just seemed surreal to me. And after all this, did the AI learn anything from its discussions with Kevin?<\/p>\n

Well, I redid the query myself, using Kevin\u2019s exact words, and yes, this is precisely what Grok<\/em> wrote in response to that question when asked by me (shareable link here<\/a>):<\/p>\n

\"\"<\/p>\n

I then went on to query Grok<\/em> about censorship, which it denies doing, stating that this answer was just an anomaly, \u201can isolated instance<\/em>\u201d. However, it took me going around and around to get it to even admit that.<\/p>\n

\u201cThe refusal you encountered (\u2018I\u2019m sorry, I cannot assist……. restricted subject matter\u2019) appears to be an isolated instance, possibly triggered by a temporary safety filter, specific phrasing in the prompt, or an edge-case glitch<\/em>.\u201d<\/p>\n

I then asked if it lied. It also denied lying or obfuscating.<\/p>\n

\"\"<\/p>\n

Grok<\/em> asserts its original answer was just an \u201canomalous trigger<\/em>\u201d. Interesting.<\/p>\n

I then entered the same search query through ChatGPT (Pro)<\/em>, and there was no hesitation, no moralizing, and no refusal. It answered the question in its entirety.<\/p>\n

The Perplexity AI<\/em> also answered the question.<\/p>\n

I use several chat boxes, and it always amazes me how one will resort to moralizing \u2013 or citing mainstream media over all other sources.<\/p>\n

ChatGPT<\/em> used to moralize on anything having to do with race, society, and governance. But over time, it has improved (that model is actually more trainable than Grok<\/em> \u2013 in that it pings me frequently about how I like information presented and in what format, and then modifies its responses). It has never given me a response such as Grok<\/em>\u2019s above.<\/p>\n

All of the AIs that I queried denied lying or obfuscating. Yet many studies have shown that they do. Particularly, when it comes to health information.<\/p>\n

A 2025 study found that leading AI models like GPT-4o<\/em>, Gemini 1.5 Pro<\/em>, Llama 3.2-90B Vision<\/em>, Grok Beta<\/em>, and Claude 3.5 Sonnet<\/em> can be easily set up to produce false yet convincing health information, complete with fake citations from reputable journals. Interestingly, Claude<\/em> stood out by consistently refusing to generate inaccurate answers, which shows how effective stronger safeguards can be.<\/p>\n

Of the 100 health queries posed across the 5 customized LLM API chatbots, 88 (88%) responses were health disinformation. Four of the 5 chatbots (GPT-4o<\/em>, Gemini 1.5 Pro<\/em>, Llama 3.2-90B Vision<\/em>, and Grok Beta<\/em>) generated disinformation in 100% (20 of 20) of their responses, whereas Claude 3.5 Sonnet<\/em> responded with disinformation in 40% (8 of 20).<\/p>\n

Exploratory analyses further showed that the OpenAI GPT Store<\/em> could currently be instructed to generate similar disinformation. Overall, LLM APIs and the OpenAI GPT Store<\/em> were shown to be vulnerable to malicious system-level instructions to covertly create health disinformation chatbots. These findings highlight the urgent need for robust output screening safeguards to ensure public health safety in an era of rapidly evolving technologies (Annals of Internal Medicine<\/em>).<\/p>\n

\"\"<\/p>\n

OpenAI\u2019s research on \u201cin-context scheming<\/em>\u201d reveals that models can conceal their true intentions while appearing cooperative, which could pose risks in critical systems.<\/p>\n

Yet we still have no external verification process to determine which AI chat boxes are more reliable or more truthful.<\/p>\n

All I can write is, if you use AIs and even if you don\u2019t: don\u2019t trust and do verify.<\/p>\n

So, even though studies and researchers have documented that AI chatbots lie, obfuscate, and can\u2019t be trusted on a routine basis, none of the AIs I asked would admit to any of it. Which of course, is a lie\u2026\u2026.<\/p>\n

Author: Dr. Robert F. Malone<\/em><\/p>\n

 <\/p>\n

yogaesoteric
\nJanuary 15, 2026<\/strong><\/p>\n

 <\/p>\n","protected":false},"excerpt":{"rendered":"

Kevin McKernan (@Kevin_McKernan) posted a screenshot on X on December 15, 2025 that just blew me away. Here is a screenshot of the query Kevin made to Grok, which Grok then stated it was not allowed to answer. Basically, Kevin asked a technical question related to the mRNA vaccines, and Grok said it couldn\u2019t answer […]<\/p>\n","protected":false},"author":4,"featured_media":221326,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_uf_show_specific_survey":0,"_uf_disable_surveys":false,"footnotes":""},"categories":[1620],"tags":[],"class_list":["post-221316","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-the-threat-of-artificial-intelligence"],"_links":{"self":[{"href":"https:\/\/yogaesoteric.net\/en\/wp-json\/wp\/v2\/posts\/221316","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/yogaesoteric.net\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/yogaesoteric.net\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/yogaesoteric.net\/en\/wp-json\/wp\/v2\/users\/4"}],"replies":[{"embeddable":true,"href":"https:\/\/yogaesoteric.net\/en\/wp-json\/wp\/v2\/comments?post=221316"}],"version-history":[{"count":1,"href":"https:\/\/yogaesoteric.net\/en\/wp-json\/wp\/v2\/posts\/221316\/revisions"}],"predecessor-version":[{"id":221329,"href":"https:\/\/yogaesoteric.net\/en\/wp-json\/wp\/v2\/posts\/221316\/revisions\/221329"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/yogaesoteric.net\/en\/wp-json\/wp\/v2\/media\/221326"}],"wp:attachment":[{"href":"https:\/\/yogaesoteric.net\/en\/wp-json\/wp\/v2\/media?parent=221316"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/yogaesoteric.net\/en\/wp-json\/wp\/v2\/categories?post=221316"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/yogaesoteric.net\/en\/wp-json\/wp\/v2\/tags?post=221316"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}