‘Perfectly real' deepfake videos likely publicly accessible within six months

Facial recognition systems are coming under scrutiny for producing inaccurate results.

A couple days after demonstrating his deepfake technology at an MIT conference on Wednesday, Hao Li affirmed to CNBC that ‘perfectly real' and ‘virtually undetectable' deepfakes are as little as six months away.

According to the deepfake pioneer Hao Li, who spoke with CNBC on Friday, ‘perfectly real' deepfake videos will be accessible to the public in less than a year. 

This estimate comes two days after having his less-than-perfect 'Vladimir Putin' deepfake technology demonstrated at an MIT conference earlier last week. Despite his video of the Russian president being clearly manipulated, the computer scientist confidently told the news media that, "Soon, it's going to get to the point where there is no way that we can actually detect [deepfakes] anymore, so we have to look at other types of solutions."

In fact, Li offered CNBC an ambitious timeline predicting that the software and tools to render unrecognizably deepfake videos could be available to the public in as little as six months to a year; such a forecast raises concerns across the world as North American political elections draw near.