Stamatis Gatirdakis, co-founder and president of the Ethikon Institute, still remembers the first time he used ChatGPT. It was the fall of 2022 and a fellow student in the Netherlands sent him the link to try it out. “It made a lot of mistakes back then, but I saw how it was improving at an incredible rate. From the very first tests, I felt that it would change the world,” he tells Kathimerini. Of course, he also identified some issues, mainly legal and ethical, that could arise early on, and last year, realizing that there was no private entity that dealt exclusively with the ethical dimension of artificial intelligence, he decided to take the initiative.
He initially turned to his friends, young lawyers like him, engineers and programmers with similar concerns. “In the early days, we would meet after work, discussing ideas about what we could do,” recalls Maria Voukelatou, executive director at Ethikon and lawyer specialized in technology law and IP matters. Her master’s degree, which she earned in the Netherlands in 2019, was on the ethics and regulatory aspects of new technologies. “At that time, the European Union’s white paper on artificial intelligence had just been released, which was a first, hesitant step. But even though technology is changing rapidly, the basic ethical dilemmas and how we legislate remain constant. The issue is managing to balance innovation with citizen protection,” she explains.
Together with three other Greeks (Apostolos Spanos, Michael Manis and Nikos Vadivoulis), they made up the institute’s founding team, and sought out colleagues abroad with experience in these issues. Thus, Ethikon was created – a nonprofit company that does not provide legal services, but implements educational, research and social awareness actions on artificial intelligence.
Copyrights
One of the first issues they addressed was copyrights. “In order not to stop the progress of technology, exceptions were initially made so that these models of productive artificial intelligence could use online content for educational purposes, without citing the source or compensating the creators,” explains Gatirdakis, adding that this resulted in copyrights being sidelined. “The battle between creators and the big tech giants has been lost. But because companies don’t want them against them, they have started making commercial agreements, whereby every time their data is used to produce answers, they receive percentages on a calculated model.”
Beyond compensation, another key question arises: Who is ultimately the creator of a work produced through artificial intelligence? “There are already conflicting court decisions. In the US, they argue that artificial intelligence cannot produce an ‘original’ work and that the work belongs to the search engine companies,” says Voukelatou. A typical example is the comic book, ‘Zarya of the Dawn,’ authored by artist and artificial intelligence (AI) consultant Kris Kashtanova, with images generated through the AI platform Midjourney. The US Copyright Office rejected the copyright application for the images in her book when it learned that they were created exclusively by artificial intelligence. On the contrary, in China, in corresponding cases, they ruled that because the user gives the exact instructions, he or she is the creator.
Personal data
Another crucial issue is the protection of personal data. “When we upload notes or files, what happens to all this content? Does the algorithm learn from them? Does it use them elsewhere? Presumably not, but there are still no safeguards. There is no case law, nor a clear regulatory framework,” says Voukelatou, who mentions the loopholes that companies exploit to overcome obstacles with personal data. “Like the application that transforms your image into a cartoon by the famous Studio Ghibli. Millions of users gave consent for their image to be processed and so this data entered the systems and trained the models. If a similar image is subsequently produced, it no longer belongs to the person who first uploaded it. And this part is legally unregulated.”
The problem, they explain, is that the development of these technologies is mainly taking place in the United States and China, which means that Europe remains on the sidelines of a meaningful discussion. The EU regulation on artificial intelligence (AI Act), first presented in the summer of 2024, is the first serious attempt to set a regulatory framework. Members of Ethikon participated in the consultation of the regulation and specifically focused on the categorization of artificial intelligence applications based on the level of risk. “We supported with examples the prohibition of practices such as ‘social scoring’ adopted by China, where citizens are evaluated in real time through surveillance cameras. This approach was incorporated and the regulation explicitly prohibits such practices,” says Gatirdakis, who participated in the consultation.
“The final text sets obligations and rules. It also provides for strict fines depending on turnover. However, we are in a transition period and we are all waiting for further guidelines from the European Union. It is assumed that it will be fully implemented in the summer of 2026. However, there are already delays in the timetable and in the establishment of the supervisory authorities,” the two experts said.
The team’s activities
Beyond consultation, the Ethikon team is already developing a series of actions to raise awareness among users, whether they are business executives or students growing up with artificial intelligence. The team’s executives created a comic inspired by the Antikythera Mechanism that explains in a simple way the possibilities but also the dangers of this new technology. They also developed a generative AI engine based exclusively on sources from scientific libraries – however, its use is expensive and they are currently limiting it to pilot educational actions. They recently organized a conference in collaboration with the Laskaridis Foundation and published an academic article on March 29 exploring the legal framework for strengthening of copyright.
In the article, titled “Who Owns the Output? Bridging Law and Technology in LLMs Attribution,” they analyze, among other things, the specific tools and techniques that allow the detection of content generated by artificial intelligence and its connection to the data used to train the model or the user who created it. “For example, a digital signature can be embedded in texts, images or videos generated by AI, invisible to the user, but recognizable with specific tools,” they explain.
The Ethikon team has already begun writing a second – more technical – academic article, while closely monitoring technological developments internationally. “In 2026, we believe that we will be much more concerned with the energy and environmental footprint of artificial intelligence,” says Gatirdakis. “Training and operating models requires enormous computing power, resulting in excessively high energy and water consumption for cooling data centers. The concern is not only technical or academic – it touches the core of the ethical development of artificial intelligence. How do we balance innovation with sustainability.” At the same time, he explains, serious issues of truth management and security have already arisen. “We are entering a period where we will not be able to easily distinguish whether what we see or hear is real or fabricated,” he continues.
In some countries, the adoption of technology is happening at breakneck speed. In the United Arab Emirates, an artificial intelligence system has been developed that drafts laws and monitors the implementation of laws. At the same time, OpenAI announced a partnership with the iPhone designer to launch a new device that integrates artificial intelligence with voice, visual and personal interaction in late 2026. “A new era seems to be approaching, in which artificial intelligence will be present not only on our screens but also in the natural environment.”
