Listen to the audio version of the article
With Ai’s new tools able to perfectly paraphrase entire texts, it becomes increasingly difficult to understand if a content is plagiarism reworked by artificial intelligence or human content. A practice that casts a menacing shadow on the violation of copyright and that already today registers some cases in journalism. This was revealed by the new report by NewsGuard, an American startup that deals with countering the spread of fake news, highlighting how the use of artificial intelligence to rewrite the contents of newspapers is spreading among low-quality sites (the so-called content farms ) who cleverly exploit Google to maximize their presence by guaranteeing themselves higher advertising revenue.
In particular, the NewsGuard report identified 37 international sites that appear to have exploited chatbots to revise articles originally published in authoritative publications such as CNN, the New York Times and Reuters, without ever citing the sources and, in fact, cannibalizing the work of editors and reporters. The sites found by NewsGuard are likely the tip of the iceberg. In fact, the sites examined have all unwittingly revealed a clue denouncing the manipulations of AI, with alert messages typical of texts generated by artificial thought, such as: “As a linguistic model of artificial intelligence I cannot rewrite this title…” or “Sorry , as a linguistic model of artificial intelligence I cannot pinpoint the content that needs to be rewritten without any context or information…”.
Copyright infringement or original content?
On the legal front, the challenge between publishers and AI is unequal, as explained to NewsGuard by Amir Tayrani, partner of the Gibson Dunn law firm, specializing in digital law. «Today it is increasingly difficult to distinguish between human-produced content and AI-generated content, and it is difficult to identify this type of potential plagiarism. At a legal level, it is not yet clear whether articles rewritten with Ai can constitute “original content”. At best, one could speak of “efficient aggregation”; at worst, of “hyper-efficient plagiarism”». While the courts, presumably, will give form and substance to the violation, the AI continues to give new form to the articles, creating a game of mirrors intended to confuse users. Google’s Bard and OpenAI’s ChatGPT usage policies, however, are clear. Google prohibits the misleading use of AI-generated content, while those of OpenAI explicitly prohibit “plagiarism”, without providing a precise definition. In Italy, the Cassation ruled in 2018 that the plagiarism of someone else’s work case of counterfeiting of the protected work, but also in the case of the so-called “evolutionary plagiarism”, as in the case of the reworkings of the Ai. With the result that the owner of the original work could then ask for the payment of the reproduction rights and compensation for damages.
The remedies for the study: the digital sticker
In the United States, steps are being taken to contain disinformation and the other dangers posed by the spread of generative artificial intelligence tools, such as copyright infringement. In July of this year, the most important companies in the sector undertook with the US presidency, albeit in a non-binding manner, to develop systems for creating a non-erasable watermark, which is automatically imprinted on all newly generated content, and other tools to make it clear which content is artificially produced and which is not. A battle in the digital world that is trying to find new ways to preserve the integrity of information and respect the boundaries of human creativity.