We live in a time where people want to believe the fakest of news. So the technology does not have to be perfect to be effective. Just the fact that fake news exists allows people to cry out “Fake News!” So the fact that deepfakes are even possible allows people to call any evidence against them fake.
A technology using machine learning and heuristic algorithms to superimpose one individual’s face onto another, frame by frame. At the time of writing, deepfake technology works best on faces in particular, but with time it will iterate and improve to cover
These individuals do not exist. These images are synthesised using GANs. While the technology is technically different to deepfakes, it demonstrates the ability of software to “imagine” highly detailed images.
What’s the most likely outcome for deepfakes?
South Africans have perhaps not been exposed to the persuasive power of deepfakes yet. However; South Africa has proven that there is a benefit to being underestimated. Our country capitalised on the extra time we were afforded before the coronavirus made an impact. There will certainly be more crises enabled in part by deepfake technology, but odds are that it happens elsewhere in the world first, thereby giving our lawmakers an opportunity to catch up with the technology.
Trust in journalism may falter if individuals only seek visual media as proof of events.
I believe the most positive outcome would be one where, like with photoshop, the public is aware enough of its functionality and persuasive power that they won’t fall for obvious attempts to manipulate. This would work for the general populace; but the protection of likenesses are going to need to be addressed on any forum where visual media can be posted, to prevent unconsented usage of another’s likenesses.
SECURANCE – The power of IT and the Law combined.
Written By Stephen Havenga