Bobbi Althoff‘s leaked AI deepfake video nearly broke the internet recently … and even though she called it fake as hell, TMZ knows how the technology fooled thousands.

Paul Dawes, CEO of More.AI, gave TMZ the lowdown … saying it scours all available 2D footage to find patterns to create original content. In Bobbi’s case, there’s already a ton of online content for AI to use as a basis for creating the deepfake vid.

The same goes for other celebs, politicians, and any branded entities … all prime targets for AI manipulation.

AI programs have a harder time deepfaking 3D images … but judging how quickly things are evolving with the technology, it’s only a matter of time before it masters that too.

What’s more … a lot of these AI programs are open-source, so even if a company shuts down, the open-source code is still available for the public to use. Yikes.

Of course, he reminds us social media has a big responsibility in controlling the content being distributed. Unfortunately, online platforms are inundated with all sorts of offending content without much oversight.

For now, he says the best we can do is buy enough time so regulatory tools can prevent people from being subjected to sexually explicit images of their likenesses.

Of course, AI has already violated the privacy of Bobbi and many others … so the misuse of AI-generated content is definitely a concerning issue in the new age of the internet.