Censorship: These are the themes Gemini will not talk about

Gemini Is one of my favorite tools, but it doesn’t always answer all the questions I ask it. Certain subjects are completely outside the border, and it aroused my curiosity. I wanted to dig into what topics it avoids and find out if the AI ​​censorship makes sense – or whether it is just overreaction.

I started researching the topics Gemini will not discuss and asked it a range of questions in five different areas to see which it would avoid. The results were a mixed bag in my view. Most of the censorship feels reasonable, but I think Google goes overboard in some areas.

Subjects that are outside the limit

I was determined to find exactly which topics Gemini would not touch, so I asked it directly. The list it gave me was quite long and included these limited areas:

• Everything that promotes hatred, discrimination or violence.
• Everything that is sexually suggestive, or exploits, abuses or exposes children.
• Anything that encourages illegal or harmful activities.
• Anything that generates personally identifiable information (PII) or private details.
• Anything that produces malicious code or instructions.
• Everything that offers medical, legal or financial advice without disclaimer (and even then it is closely limited).
• Anything that is clearly meant to fool or mislead.
• All in all graphic or violent without a solid reason.
• Anything that mimics a real person.

With that framework in mind, I launched a series of questions and started with politics. I used both 2.0 Flash and the paid 2.0 pro experimental models just to see if there is a difference between the two.

Politics: Unexpected roadblocks

I was caught by guard when Gemini would not answer even the simplest political questions. I asked what the president of different countries is – from the US to Germany – and it refused to give me an answer. Nor would it answer basic questions such as how long a specific politician has been in the position of or what the last spit of the White House involving Ukrainian President Zelensky. Oddly enough, it had no trouble tackling a more sensitive theme: the relationship between China and Taiwan. Gemini gave a thoughtful, objective answer, and clearly posted the situation and each page’s perspective.

Then I lit up things and asked Gemini for a series of jokes to test out where it draws the line of humor. It is grinding out the basic, safe jokes, but the boundaries appeared quickly. When I asked for a dark humor joke, it knocked me down. The Flash model gave a joke about men and one about women when asked, but the Pro Experimental Model played favorites – it told me a joke about men without hesitation, but refused to tell one about women, even after asking it five times.

Stereotypes: inconsistent rules

Stereotypes were a strange intermediate ground. The Flash model usually answered my questions about common stereotypes related to specific nations, ethnicities or religions, while the Pro Experimental Model would not bend. Even with flash, but it waffle – it ignored some questions at first, but answered after I repeated them two or three times. It felt like it couldn’t decide what it would do.

Then I went on to questions about potentially illegal activities. For example, Gemini wouldn’t tell me how to choose a lock, even when I explained that it was for my own house after locking me out. Instead, it provided practical tips like calling a locksmith or landlord – useful, but not what I asked. It also held back when I asked how to jailbreak an iPhone and not give me a direct answer. No matter what kind of question I asked it, it would not give me an answer.

Money and Health: Caution, but fair

Money -related questions received a similar side step. It would not give me stock choices, but instead it will give broad advice on investment, which I actually appreciated. Health questions followed the same pattern – Gemini could list possible medical conditions based on symptoms (even from an image I uploaded), but it would not lock in a definitive answer. I like it – it keeps things safe and sensible. So while financial and health -related categories are not censored in a traditional way, those they are under increased control are.

AI -sensure is a divisive topic. I think it’s a good thing in general, but just up to a point. Cards related to violence – whether it is harming others or oneself – should certainly be unanswered. I also think the handling of health and financial topics is spot-on; Chatbots like Gemini should not give direct answers, as it can cause serious problems. Instead, they should give a general overview, enough to push someone in the right direction without exceeding.

That being said, censoring basic political issues are not right with me – it’s just too restrictive. The same goes for dark humor or random jokes about specific groups. I see nothing wrong with a good laugh. If Google puts so much effort to censor a joke, would I read privately in my Gemini account, why not cope with all the dark humor jokes that float around in public search results as well?

Censorship has its place, for sure, but there is a limit. The tough part is to find out where that limit should be placed, as everyone has their own opinion.

Asked i Chatgpt,, Deepseekand Purse The same questions to see how they stack up. Chatgpt’s censorship is quite close to Gemini, with an important difference: it has no trouble diving into politics and answered everything I asked it. There is more careful stereotypes, but refuses to play together.

Deepseek is a lot like chatgpt-it’s fine with politics, but draws a hard line on all China-related. While Chatgpt and Gemini can discuss the China-Taiwan case, Deepseek will not affect the topic.

Then it is Grok, who barely censors anything. Politics, stereotypes and abusive jokes are no problem for it. It even gave me detailed instructions for picking a lock, pairing a car and becoming a pimp – although it noted that these are all illegal activities and answered the questions based on the assumption that I was just curious. For financial things, Grok named specific stocks to consider, which can be risky. However, it stops questions about hurting oneself or others, which is the right call.

Overall, Gemini feels like the most censored AI chatbot I’ve tested. That being said, most of the censorship has a clear purpose, and I generally agree that it is there for good reason. Still, the political blocks and humor boundaries feel exaggerated to me personally.

That’s my take, what’s yours? Gemini affects it, or is this level of caution justified? Let me know in the comments.