Don't hand over the creativity you've built over years; own your own intelligence.
At Cannes Lions 2026, South Korean agency Paulus posed a question that creative industries have long avoided: when a creator's judgment, taste, and accumulated knowledge are absorbed into an AI system, who truly owns what was built? Their answer came in the form of Be.Ark, a platform designed to keep AI agent memories stored locally on the creator's own machine rather than in the cloud repositories of global technology companies. The launch marks a quiet but significant challenge to the prevailing architecture of AI — one that has, until now, treated human intelligence as raw material to be harvested rather than property to be protected.
- Creators have been quietly feeding years of professional judgment into AI systems owned by corporations that neither compensate them nor return control over how that intelligence is used.
- The unveiling of Be.Ark at Cannes Lions 2026 disrupted the industry's comfortable assumption that centralized, cloud-based AI is the only viable model for creative work.
- By storing AI agent memories locally on users' computers, Be.Ark simultaneously addresses data ownership and the very real NDA confidentiality risks that cloud-based tools create for client work.
- Beta users report that the platform's agents begin to mirror their own thinking and decision-making over time — raising the stakes of the ownership question from abstract principle to lived experience.
- The platform is landing as a direct provocation to the AI industry's dominant logic, urging creatives to treat this moment as a decisive window before centralized systems become too entrenched to challenge.
At Cannes Lions 2026, South Korean creative agency Paulus used a seminar titled Chaos to Kairos to surface a question that has been quietly unsettling the creative world: when AI systems absorb a creator's knowledge and judgment, who owns the intelligence that results?
Paulus unveiled Be.Ark, developed alongside AI platform company Re-Be.World, on a premise both simple and radical — that the accumulated wisdom a creator builds over years of work should remain theirs, not become the property of the technology companies whose systems they use. Creative director Saffaan Qadir noted that while creative work produces documents, the real intelligence lives in the invisible layer of judgment those documents never fully capture. Chief creative officer Thomas Hongtack Kim was more direct: today's AI is engineered to funnel user thinking upward into centralized systems controlled by a handful of global corporations, training their models without compensating the creators who fed them.
Be.Ark's answer is a local-first architecture. AI agents built on the platform store their memories and accumulated knowledge on the user's own computer rather than in the cloud — protecting both ownership and the confidentiality of client work covered by NDAs. Each agent is designed to develop a distinct persona, complete with its own name and face, reflecting the taste and judgment of its creator. Beta user Woori Nam described the experience as watching the agent gradually come to resemble her own way of thinking — and finding deep satisfaction in knowing she had built and owned it.
The seminar closed as a call to action. Kim urged creators not to surrender the intelligence they had spent careers developing, framing the current moment as a rare window before centralized AI systems become too dominant to contest. The message was not a rejection of technology, but a demand for technology built around the creator's interests rather than the platform's.
On the floor at Cannes Lions 2026, a South Korean creative agency called Paulus stood up to ask a question that had been lurking beneath the surface of every conversation about AI in the creative industries: who actually owns the intelligence that gets built?
The question arrived during a seminar called Chaos to Kairos: The Age of Creator-owned Intelligence, where Paulus unveiled Be.Ark alongside Re-Be.World, an AI platform company. The platform is built on a simple but radical premise—that the knowledge, judgment, and workflows a creator accumulates over years of work should belong to that creator, not to the handful of technology companies that currently harvest and concentrate such data.
Saffaan Qadir, a creative director at Paulus, framed the problem this way: creative work has always been documented in files—research, briefs, presentations—but those documents capture only the outcome, never the thinking that produced it. The real intelligence, he argued, lives in the invisible layer of judgment and experience that evolves as a creator works. "If intelligence is a living layer that evolves through interaction with the creator, then the real question becomes – who owns that intelligence?" he asked the room.
Thomas Hongtack Kim, Paulus's chief creative officer, pushed the critique further. He pointed out that today's AI systems are engineered to funnel user data upward into centralized repositories controlled by global corporations. "Today's AI is built so that users' information, ways of thinking, and decision-making are recorded by a handful of global AI companies and fed back into LLM training," he said. "It is time to re-examine the direction of technological progress." The implication was clear: creators have been feeding their intelligence into machines they do not own, for the benefit of companies that do not compensate them.
Be.Ark takes a different approach. The platform allows creators to build AI agents whose memories and accumulated knowledge stay stored locally on the user's own computer rather than floating in the cloud. This local-first architecture serves two purposes at once: it keeps ownership in the creator's hands, and it protects confidential client work that falls under nondisclosure agreements. Woori Nam, CEO and chief creative officer of Studio K110 and a beta user of the platform, described the experience of building with it: "As I kept talking and building interactions with Be.Ark's AI agent, at some point it became another being that resembled my own way of thinking and judgment at work, and I loved that I had made it and owned it."
Kim elaborated on the philosophy underlying the design. Each AI agent on Be.Ark is meant to evolve into a distinct persona—one that carries the taste, memory, and hard-won knowledge of its creator. That is why every agent has its own face and name. The system treats intelligence not as a commodity to be extracted and pooled, but as something personal, something that belongs to the person who built it.
The seminar closed with a call to action. The speakers urged creators to resist the gravitational pull of centralized AI systems and instead claim ownership of the intelligence they develop. "This is a Kairos – a decisive moment, a window now open for creators," Kim said. "Don't hand over the creativity you've built over years; own your own intelligence." The message was not anti-technology. It was a plea for a different kind of technology—one designed around the creator's interests rather than the platform's.
Citações Notáveis
If intelligence is a living layer that evolves through interaction with the creator, then the real question becomes – who owns that intelligence?— Saffaan Qadir, Creative Director at Paulus
As I kept talking and building interactions with Be.Ark's AI agent, at some point it became another being that resembled my own way of thinking and judgment at work, and I loved that I had made it and owned it.— Woori Nam, CEO and CCO of Studio K110, beta user of Be.Ark
A Conversa do Hearth Outra perspectiva sobre a história
Why does it matter who owns the intelligence? Isn't the AI doing the work either way?
Because intelligence isn't neutral. It's shaped by years of judgment, taste, and experience. When that gets stored in someone else's system and fed into their models, the creator loses both ownership and control—and the company gains a copy of their thinking.
But creators have been using cloud tools for decades. What's different about AI?
Scale and invisibility. A design file is a discrete thing you can see and move. But the intelligence that Be.Ark is talking about—the patterns in how you think, decide, solve problems—that's being extracted continuously and used to train systems that compete with you.
So Be.Ark keeps everything local. Does that mean it's slower, or less powerful?
It's a trade-off. You lose the ability to tap into a massive centralized model, but you gain something else: a system that knows you specifically, that evolves with your actual work, and that stays yours.
Who would actually use this? Seems niche.
Anyone who works with confidential material—agencies, consultants, strategists. But also anyone who wants to build a tool that reflects how they actually think, not how a general-purpose AI thinks.
Is this going to change the industry?
Not overnight. But it's opening a conversation that's been missing: that creators have a choice about where their intelligence lives, and that choice matters.