If it is using chatgpt as a backend, my guess is that they are using Azure OpenAI and know what they are doing.

Azure OpenAI allows you to turn off abuse monitoring and content filtering if you have legitimate reasons to do so.

It would be very hard for a malicious actor to get the approval to turn off both using a front company. But if one would manage to do it, they could create such malicious chatGPT service with little to no chance to be found out.