(By Alessio Jacona*) British and
global politics have had a new taste of the risks to democracy
represented by so-called deepfake, namely fake multimedia
content created using artificial intelligence (AI).
During the annual Labour party conference in the UK, accounts on
the social media platform X (formerly known as Twitter) released
a video containing audio in which party leader Sir Keir
Starmer's counterfeit AI voice can be heard swearing at and
abusing members of his staff.
The episode represented a fake and full-blown attack on the
Labour leader from the internet just as he is starting to
prepare his political campaign for the general elections that
are due to be held between the end of 2024 and early 2025.
Most of all, it seems to have been launched on purpose to
discredit him just when the polls are giving his party - which
has been in opposition for 15 years - a lead of almost 20 points
over Prime Minister Rishi Sunak's Conservatives.
The deepfake was initially published on an account with only
3,500 followers, in a post that read as follows: I have obtained
audio of Keir Starmer verbally abusing his staffers at
conference. This disgusting bully is about to become our next
PM." At the time of writing, the content had been viewed 1.5
million times. Shortly after the first video, a second piece of
content was posted in which Sir Starmer's counterfeit voice can
again be heard, this time criticising the city of Liverpool.
This is the second episode of its kind in Europe in less than
two weeks. On September 28, just two days before the
parliamentary elections in Slovakia won by the former
pro-Russian Prime Minister Robert Fico, a deepfake circulated on
Facebook in which Michal Šimečka, leader of the liberal
Progressive Slovakia party, could be heard discussing with
Monika Tódová, a journalist working for the daily newspaper
Denník N, how to rig the elections by buying the votes of the
Roma minority. The pair immediately denounced the content as
being fake, but the electoral silence imposed under Slovak law
for 48 hours before the vote (i.e. exactly when the deepfake
audio was published) made it extremely difficult to prevent the
content from being spread on Facebook.
"This is a phase in which the risk of a 'Wild West' is extremely
high because if, on the one hand, powerful and accessible AI
technologies exist and are multiplying, on the other there is
still not a widespread culture of understanding around the risks
that they can entail also for democracy, there are no AI-based
technologies capable of countering the spread of deepfakes, and,
lastly, there is still a lack of common transnational
legislation," explains Stefano Epifani, president of the
Foundation for Digital Sustainability.
The foundation presided over by Epifani has recently released a
Manifesto for the Digital Sustainability of AI, a document that
grew out of the collaboration between philosophers, lawyers,
engineers and university professors in order to analyse the
challenges and opportunities presented by AI and, for each
Sustainable Development Goal (SDG), identify those
characteristics of AI that contribute more than others to their
achievement.
Epifani said on order to counter the phenomenon of deep fakes it
is necessary "to create a widespread awareness of the risks and
opportunities of AI, but also new legislation that obliges the
platforms that disseminate fake content to develop and use
AI-based countermeasures in order to guarantee users the truth
of what they enjoy".
*Journalist, innovation expert and curator of the ANSA.it
Artificial Intelligence Observatory
ALL RIGHTS RESERVED © Copyright ANSA