Published in Athens Voice on 27 September 2024
Artificial Intelligence Content Authenticity is the central subject of this article, originally published in Athens Voice, examining the growing legal concerns surrounding AI-generated images, audio and video.
It is a fact that an increasing amount of artificial intelligence-generated content is appearing online every day. Artificial intelligence is a powerful tool that has been used in many different ways by major corporations, influencers, web developers, advertising companies, mobile telecommunications companies, as well as ordinary citizens for entertainment purposes.
But what happens when this content has been created to insult someone or, even worse, to incriminate an innocent person — particularly when this false yet realistic creation results in that person being brought before the courts as the alleged “perpetrator”?
Numerous advertisements have appeared that are products of artificial intelligence — some realistic and others less so — with the sole purpose of extracting money from members of the public.
Well-known doctors and nutritionists appear to “present” miraculous dietary supplements, while actors, singers, influencers and other widely recognised individuals appear to promote trading platforms, online betting companies and other products or services.
When someone falls victim to such a scam and, for example, suffers damage to their health as a result of an illegal or unsafe pharmaceutical product, it is highly likely that the case will proceed to court.
The victim may take legal action against the nutritionist or doctor who appeared to be “selling” the product, while the doctor or nutritionist — who has also fallen victim to the fraud — may in turn take legal action against the person who created the fake advertisement using their identity.
In cases such as this, the innocence of the doctor or nutritionist concerned can be relatively easily verified. The payment account — or electronic payment platform, such as PayPal — will be registered under another name, the IP address from which the advertisement was uploaded will not belong to them, and there may be several other pieces of evidence that can help establish their innocence.
Artificial Intelligence: Content Authenticity and Legal Concerns
But what happens when a video or audio recording is presented before a court or the authorities as evidence of guilt?
A few days ago, I was sitting in a café in central Athens waiting for a friend. I unintentionally overheard the conversation taking place at the table directly next to mine.
A man who was in the process of divorcing his wife and determining the custody arrangements for their children was discussing the proceedings with another person.
At one point, he said that because he was afraid he might not be granted custody, he intended to use artificial intelligence to create an audio recording using his wife’s voice, in which she would appear to speak badly and threateningly towards their children. He would then present the recording in court, claiming that she was unfit to have custody.
Other audio “evidence” or videos could be created in the same way with the intention of incriminating people for virtually any act — and yes, this is not impossible.
Given the level technology has now reached and the rate at which it continues to develop, such incidents are likely to become increasingly common.
In the example above, we are not dealing with financial fraud that could potentially be identified through a payment account or IP address, nor are we dealing with the sale of products.
Instead, we have an artificial intelligence-generated creation with potentially extremely serious consequences for the woman who is the victim, and material whose authenticity may be difficult to determine conclusively.
In any case, actions of this kind can also occur in more ordinary, everyday personal situations.
A malicious individual could place your photograph next to that of another woman, make it appear that you were present in a location that conflicts with your beliefs or values, make it appear that you were speaking badly about a friend, or create any number of other imaginative but entirely false scenarios.
Is it possible to prove that evidence of this kind was created using artificial intelligence?
Most of the videos and audio recordings currently circulating online may be of good quality, but they often contain elements that reveal that they are not genuine.
There are, however, “professionals” in this field who are capable of creating audio recordings or videos that appear completely realistic.
Software designed to assess the authenticity of such material is available on the market, although these programs cannot determine with 100% certainty whether something was created using AI.
In any case, when an audio recording, video or photograph is presented in court or to the authorities, it may potentially be identified as artificial by specialised personnel from the Cyber Crime Division.
However, we should consider how much more time-consuming judicial proceedings could become if a hearing has to be adjourned so that the evidence can undergo expert examination.
In the United States, this phenomenon has already begun to create similar problems. Not only are AI-generated materials being presented in court as evidence of innocence or guilt, but parties may also falsely claim that genuine photographs and videos presented by the opposing side were generated by AI.
As a result, proceedings may be postponed so that the material can be examined by an expert, making the entire process increasingly complex and time-consuming.
Manipulation and defamation have always existed
The difference today lies in how difficult such manipulation can be to identify, how easily these tools can be accessed by the public and the sheer number of programs that are now available.
Since the early days of the internet, photographs have been manipulated using Photoshop, while videos and audio recordings have been altered through editing and specialised software.
However, the quality available today is entirely different, as is the speed with which such material can now be created, compared with the countless hours that would have been required in previous years to create or improve a video or photograph.
These new factors resulting from artificial intelligence have created disruption and concern across a variety of sectors — not only among judicial or police authorities, but also among ordinary citizens who fear becoming victims of similar forms of fraud.
What is certain is that we have not yet encountered all the problems that artificial intelligence may create. Something new emerges almost every day, and we do not yet know what the ultimate capabilities of this technology will be.
If someone had told us ten years ago that today we would have such powerful tools in our hands, it would have seemed unrealistic.
Therefore, no one can know what may emerge over the next ten years.
