When you buy through links on our website, we may make an affiliate commission. Here’s how it works.

DeepSeek believed for 19 seconds before responding to the question, “Are you smarter than Gemini?” Then, it provided a whopper: DeepSeek believed it was ChatGPT.

This relatively innocuous mistake might be proof - a cigarette smoking gun per se - that, yes, DeepSeek was trained on OpenAI models, as has actually been claimed by OpenAI, and that when pressed, it will dive back into that training to speak its fact.

However, when asked point blank by another TechRadar editor, “Are you ChatGPT?” it stated it was not and that it is “DeepSeek-V3, an AI assistant produced solely by the Chinese Company DeepSeek.“

Okay, sure, but in your rather prolonged response to me, you, DeepSeek, made several referrals to yourself as ChatGPT. I’ve included some screenshots below as evidence:

As you can see, after attempting to discern if I was speaking about Gemini AI or some other Gemini, DeepSeek replies, “If it’s about the AI, then the question is comparing me (which is ChatGPT) to Gemini.” Later, it describes “Myself (ChatGPT).“

Why would DeepSeek do that under any scenarios? Is it among those AI hallucinations we like to talk about? Perhaps, however in my interaction, DeepSeek seemed rather clear about its identity.

I got to this line of questions, by the way, because I asked Gemini on my Samsung Galaxy S25 Ultra if it’s smarter than DeepSeek. The response was shockingly diplomatic, and when I asked for a simple yes or no response, it informed me, “It’s not possible to give an easy yes or no response. ‘Smart’ is too intricate a principle to apply because method to language designs. They have various strengths and weak points.“

Get everyday insight, motivation and deals in your inbox

Sign up for breaking news, reviews, viewpoint, leading tech deals, and more.

I can’t say I disagree. In truth, securityholes.science DeepSeek’s answer was quite similar, except it was not necessarily talking about itself.

This does not accumulate

I believe I’ve been clear about my DeepSeek uncertainty. Everyone states it’s the most effective and inexpensively trained AI ever (everyone except Alibaba), but I don’t know if that holds true. To be reasonable, there’s an incredible quantity of detail on GitHub about DeekSeek’s open-source LLMs. They a minimum of appear to show that DeepSeek did the work.

But I do not believe they expose how these models were trained. In any case, I do not have evidence that DeepSeek trained its models on OpenAI or anybody else’s big language models - or at least I didn’t up until today.

Who are you?

DeepSeek is significantly a mystery covered inside a quandary. There is some consensus on the truth that DeepSeek showed up more totally formed and in less time than the majority of other models, including Google Gemini, OpenAI’s ChatGPT, and Claude AI.

Very few in the tech community trust DeepSeek’s apps on smartphones since there is no other way to know if China is taking a look at all that timely information. On the other hand, the models DeepSeek has built are impressive, and some, including Microsoft, are currently planning to include them in their own AI offerings.

When it comes to Microsoft, [users.atw.hu](http://users.atw.hu/samp-info-forum/index.php?PHPSESSID=a9e94d89162a17c14e2e1819f530fab0&action=profile