I'll verify for you: it's not true. ChatGPT vehemently avoids making jokes about religions, but if you weasel it into doing so it will do so fairly across religions.

That being said, ChatGPT is flexible and chaotically random and so some people will statistically find examples of bias.

In the grand scheme it's impressively unbiased, especially given that ML models are implicitly biased and removing bias is the extremely difficult part. It's sad that many people misunderstand this.