Around twenty tech giants, including Google, Meta and Microsoft, have just signed an agreement defining strong principles to fight against deepfakes and other misleading content generated by AI, which threaten the integrity of global electoral processes. Let us examine the essential commitments of this historic text.
Tech giants speak out against electoral manipulation by AI
Artificial intelligence (AI) is causing a revolution across all sectors, bringing both promise and threat. Hackers have understood this well, they who divert the latest technological innovations to create weapons of mass manipulation.
According to a article published in Le Monde, the main technological players signed an unprecedented agreement on Friday to counter one of these cyber threats: electoral deepfakes generated by AI. These hyperfakes which perfectly imitate the faces, voices and gestures of political figures risk sowing doubt and disinformation in the run-up to crucial elections.
The signing took place during the Munich Security Forum, bringing together international leaders such as American Vice President Kamala Harris, President of the European Commission Ursula von der Leyen and Ukrainian President Volodymyr Zelensky.
The timing of this agreement is crucial. According to Le Monde, almost half of the world’s population of voting age will be called to the polls in 2024, including major elections such as the American presidential election.
Concretely, this three-page agreement unveiled on February 16 targets Audiovisual content generated by AI to impersonate candidates or spread false news about voting methods. A massive manipulation of minds which could undermine democratic processes, hence this call for the general mobilization of the technological sector.
A threat still limited, but growing
The signatory companies undertake to develop tools to ensure the traceability of content generated by artificial intelligence (AI). The objective: to be able to easily trace the source of creation of a misleading image or video.
This could, for example, involve integrating invisible “digital watermarks”, a kind of marking incorporated at the very moment of generation by the AI algorithm. This traceability would greatly facilitate the moderation of this sensitive content. Work in this direction is already underway, notably with the development of standards such as C2PA.
The signatories also plan to deploy technologies to automatically detect misleading content once published online. And mechanisms to make it easier for users to report suspicious content. Of course, while striking a balance between effective moderation and respect for freedom of expression.
While verified cases of electoral deepfakes still remain marginal, the risk of sophisticated cyberattacks during the 2024 elections is real and taken very seriously. Remember that many women are already victims of non-consensual deepfakes made with the same generative AI tools, such as recently the star Taylor Swift.
The agreement signed by the digital giants is therefore a good start, but does not constitute a definitive solution. Vigilance by all stakeholders in society will be necessary to maintain the integrity of the democratic process in the era of generative AI.
Maximize your Tremplin.io experience with our ‘Read to Earn’ program! For every article you read, earn points and access exclusive rewards. Sign up now and start earning benefits.
