The underlying content is largely irrelevant.

What is relevant is what is advertised, and for what purpose. In many cases this alone may be sufficient to show intent/malice/negligence, especially when it breaks law. I don't imagine OP has written consent forms from these people authorizing the use of their voice for this purpose.

This would inevitably come first before even examining content.

If the voice example isn't similar enough, then that may be grounds to indicate fraud. Either way you cut it, law is broken, and there is a lot of applicable law that touches on deepfake technologies.