Showing how easy it is to make deepfakes of politicians using artificial intelligence, independent senator David Pocock creates AI videos of Prime Minister Anthony Albanese and Opposition Leader Peter Dutton.
I won't pretend to know what the solution is but I am very grateful that he is raising the issue publicly. We need to start working on this stuff now, before it gets completely ingrained and normalised within society like algorithmic social media did.
As the article already states, even with a ban it would still be a reactive process of taking it down after-the-fact, which we already know Isn't Great™ based on other times when the juicy-but-false news headlines that come out first get more eyeballs than the later corrections.
Still, on balance it's probably better than a complete free-for-all? I guess there'd need to be clear lines about what's okay / not, as it seems very easy to overdo it if the language is too vague (e.g., in an extreme case, accidentally banning all manipulation of images / video in political contexts).
I'm not necessarily advocating it. I put the link up because its a useful addition for a post like this.
In saying that the idea that bans don't work takes the 'war on drugs/prohibition' approaches out of context.
I'm writing this from memory (of reading, not experience :p ) because i don't have time to go and reread it all so apologies if details are wrong, the essence should be there though,
Prohibition was enforced on the population by ideological puritans in power at the time. It seems no clear popular support backed or accepted the prohibition's rationale and is a driving reason why it was so hard to maintain and dropped.
'War on drugs' ideas should be dropped because the evidence shows the American public have not benefited from the policy position, and in fact the 'War on Drugs' has likely increased the costs and harms associated with the drugs trade rather than diminished them. So, while we can say the 'War on Drugs' enjoyed popular support, in contrast to Prohibtion, the health, economic, violence, and consumption patterns have all trended negative against the policy over the period, meaning the policy has failed in its stated objective and needs changing.
The point of these two examples being referred to when considering other bans isn't to sit on the ideological plane of libertarians and shout "All bans are bad, you won't tread on me." But to consider the negative implications of a proposed ban and how its reality could differ from the vision, and adjust accordingly.
There are enforced bans throughout society, think driving without a seatbelt, driving on the wrong side of the road, electrician sign offs, work with and manufacture of radioactive materials, essentially anything the enforceabke by the police and courts you can argue is a 'banned practice'.
A ban targeting political party practices is far more enforceable than population wide bans, its a smaller 'market', with known players, to regulate. I beleive Lobby groups in Aus also have to identify themselves when they put out attack ads.
All that said, if a ban was implemented it doesn't stop AI use in political advertising, but it does set the tone, and that means a lot. We as a society can't stop murders, but we can build up barriers against their use as a legitimate tool of pursuing ones goals.