Accenture Technology Vision 2022 Research Reflects on the Pitfalls of Deepfakes
A few years ago, the user who edited videos, audios or images needed to have intrinsic technical knowledge to handle these files. Today, with the spread of Artificial Intelligence (AI), more people now have access to multiple editing techniques, which has contributed to the dissemination of fraudulent news.
In practice, the term deepfake comes from a combination of deep learning (deep learning) and fake (fake), consisting of a strategy of associating human images or sounds, based on artificial intelligence techniques, to match speech or the scene from an existing recording. The result is synthetic content, but very well done, such as the speech by Ukrainian President Volodymyr Zelensky announcing the surrender of Russian troops. Or the pope dressed in haute couture. There was also former US President Donald Trump being truculently arrested by New York police.
The fact is that deepfakes are everywhere, even in the soap opera Travessia, on TV Globo, in which the protagonist of the plot, Brisa, was transformed, via AI, into a child kidnapper, thus becoming a victim of deepfake. Another recent case was the advertisement for Volkswagem, which recreated the singer Elis Regina in an advertising piece, causing discussion on the internet and becoming a case of ethical representation by Conar.
This is why the topic is so important, in the opinion of Filipe Bento, CEO of Br24, the largest producer of Bitrix24 content in the world: Today, the system for building deepfakes is so improved that the recognition of what is true (true) and what is fake can become a very difficult task. When deepfake was born, in approximately 2014, its use was more complicated, due to the requirement of advanced knowledge in technology, however, as technology advanced, easier solutions to produce deepfakes were emerging.
And companies, if they are not careful, are in the crosshairs of this practice. Potential victims are public figures, whose image or voice is widely available on the web. Once a victim, there is not much that can be done, because the main characteristic of deepfake is the fast speed of dissemination of the photo, audio or video.
The picture can be even more serious. A survey by Accenture Technology Vision 2022, entitled Meet me in the metaverse, points out that 65% of global consumers do not know how to recognize or identify deepfake videos or synthetic content. That’s why prevention, according to Filipe Bento, is the best medicine: We always recommend brand safety, which refers to a set of practices to protect companies from unwanted people and environments. Among the main measures, it is worth highlighting the use of adtech technology, including pre-checking, in which it is possible to filter in advance any suspicion of invalid traffic.
He remembers that embarrassments are not just related to reputation. On behalf of a company, hackers can actually collect information from their customers, harming people and putting the business in adverse situations under the General Data Protection Law (LGPD).
