DeepSeek AI Model Mistakenly Identifies Itself as ChatGPT, Raising Questions About AI Training

DeepSeek is facing renewed scrutiny after reports that its AI chatbot repeatedly identified itself as ChatGPT, raising questions about how artificial intelligence models are trained and whether competing AI systems rely on each other’s outputs.

According to a recent claim shared on X (formerly Twitter), a user asked DeepSeek to identify itself 30 times. In 26 responses, the chatbot reportedly claimed to be ChatGPT rather than DeepSeek.

The unusual behavior has sparked online discussions about AI model training, transparency, and the practice of using outputs from existing AI systems to develop new models.

Why Is DeepSeek Calling Itself ChatGPT?

DeepSeek, a Chinese artificial intelligence company, has developed AI models designed to compete with leading systems from OpenAI, Anthropic, and Google.

However, recent user reports suggest that DeepSeek may sometimes confuse its identity with ChatGPT.

One X user claimed to have tested the chatbot 30 times, receiving 26 responses in which DeepSeek identified itself as ChatGPT.

Separately, an online AI fingerprinting test has reportedly suggested similarities between a newer DeepSeek model and OpenAI’s GPT-6 Astra model.

These claims have attracted attention, but similarities in chatbot responses do not establish that one company copied another company’s model.

DeepSeek Has Faced Similar Questions Before

This is not the first time DeepSeek has been associated with questions about AI model identity and training data.

In 2024, reports suggested that DeepSeek V3 identified itself as ChatGPT in five out of eight tests. Some responses also reportedly resembled jokes previously generated by GPT-4, including their punchlines.

More re