When Péter Szabó speaks, it’s with quiet authority. Not from flash or flair, but from years spent studying the mathematics behind intelligence. At heart, Szabó is a researcher. He’s also the co-founder of Tengr.ai, a European company that quietly launched the world’s first AI-generated art exhibition. And he is intensely aware of what that means for privacy, artistry, and the future of human-machine collaboration.
“We were the first to put AI art into a gallery setting,” he says, matter-of-factly. “Before it went viral. Before the headlines. Before it became normal.”
Szabó’s path to generative AI didn’t begin in a research lab or with Silicon Valley bravado, it evolved from a career grounded in user experience. As the author of User Experience Mapping (Packt, 2017), he originally focused on how humans interact with complex systems. His early research career, shaped by work in probability, signal processing, and theoretical modeling, laid the groundwork for what would later become the core logic behind Tengr’s foundational model. The shift toward generative models was a natural progression driven by a deep interest in applying mathematical structures to real-world interfaces and design challenges.
“We didn’t want to just copy others. We built it from scratch. We trained our own model in Europe. That matters.”
That emphasis on sovereignty and privacy is no accident. Szabó believes that personal data, especially when connected to creative identity, must remain under the control of the individual. Tengr does not harvest user data. It doesn’t phone home to unknown servers. It is local, private, and intentional by design.
“There is a difference between convenience and trust,” he says. “We wanted artists to know exactly what happens when they use our tools. No hidden terms. No silent data collection.”
That clarity is also embedded in Tengr’s interface. One of the company’s defining principles is to never show an error message.
“We call it Hyperalign,” Szabó explains. “The idea is simple. If a person asks something that doesn’t work, the system adapts. It should never say ‘invalid input’ or ‘try again.’ That’s our job, not the user’s problem.”
Tengr’s system is designed to quietly self-correct, using real-time feedback to guide the user toward successful outputs. That philosophy makes the technology feel less like a machine and more like a creative collaborator.
Tengr’s commitment to transparency and data integrity also sets it apart on the thorny issue of copyright. While other platforms like Midjourney have come under fire for training their models on copyrighted works scraped from the web without consent, leading to high-profile lawsuits and industry backlash, Tengr’s approach is different. Its foundational model is trained using carefully curated, legally sourced datasets.
Szabó insists that respecting the rights of original artists is non-negotiable. “We are not interested in building on stolen work,” he says. “If generative AI is to have a future, it must be built on ethical ground.”
But how do artists feel about it?
Szabó doesn’t flinch at the question. He’s been asked many times whether generative AI is an insult to traditional creators.
“I think it’s the wrong question,” he says. “When painters were first expected to reproduce perfect likenesses, that was seen as skill. Then came photography. And suddenly, painters were free. That’s when you get impressionism, cubism, surrealism.”
He pauses.
“So ask yourself, is AI limiting the artist? Or is it liberating them from the burden of perfection?”
This perspective is central to Szabó’s work. He doesn’t see Tengr as replacing human creativity. He sees it as shifting the threshold. By removing the labour of technique, it allows room for new forms of experimentation and style.
“It’s not about outsourcing your art. It’s about exploring new layers of your imagination. That’s what the best artists have always done.”
The conversation turns unexpectedly toward accessibility. Szabó is passionate about disability rights and inclusion, and sees AI as a bridge rather than a barrier.
“People often forget that not everyone can hold a brush or sit at an easel. What if your body doesn’t allow it? Should you be excluded from art entirely? Or can we offer new tools to access that same joy of creation?”
This is not theoretical for him. Tengr is already working with organizations to bring generative tools into rehabilitation settings, helping users with physical disabilities express themselves in new ways.
“Creativity is not a luxury,” he says. “It is a core human need. And it should be accessible to everyone.”
Szabó’s tone remains composed, but the message is clear. Tengr is building for more than just aesthetics. It’s building for dignity.
From a product standpoint, Tengr remains proudly European. The team is based across Hungary, Germany, and Austria, with training infrastructure designed to comply with European data privacy laws. That sets it apart from competitors trained on massive datasets with murky origins and little regard for provenance.
“We believe provenance matters,” Szabó says. “Where the data comes from, who owns it, how it is used. These are not minor concerns. These are the foundations of trust.”
Despite the serious tone, Szabó retains the curiosity of a scientist. He is less interested in the hype around AI than in its structural potential. His interest lies in models that can align, adapt, and respond without needing to dominate.
“The future is not about building smarter machines. It is about building more respectful ones.”
He doesn’t see himself as a founder in the traditional tech sense. No venture capital theatrics, no big marketing campaigns. Just the steady progress of a team that believes in the work.
“I love research. That’s what keeps me going. Solving problems at the edge of what we understand. It’s not glamorous. But it’s where the breakthroughs happen.”
Tengr is already powering design workflows across Europe, with more partnerships on the horizon. The platform has surpassed 100,000 users, with an average session length of 28 minutes and 62 percent returning weekly. That is a clear signal that users are not just dabbling but actively working with the tool. Still, Szabó is careful not to overclaim. For him, success is not about market share or daily active users.
“Success is when someone uses our tools and feels seen. When they create something they didn’t think they could. That’s what I want.”
As AI continues to surge into every domain of culture and industry, Szabó’s approach offers a different template, one rooted in mathematics, humility, and the simple belief that good technology should be kind, private, and deeply human. Long before the current AI boom or tools like PyTorch existed, Szabó was already engaged in the field working without the benefit of large models or current-scale datasets underscoring his pioneering role and long-term vision.
