At WAiFF 2026, directors Claude Lelouch and Mathieu Kassovitz warn AI could reshape film faster than rules can

At the World AI Film Festival (WAiFF) 2026, French directors Claude Lelouch and Mathieu Kassovitz put blunt language to a tension already running through filmmaking: how to use artificial intelligence without washing out what makes a movie feel singular—or destabilizing the jobs that make it.

In a conversation reported by Mediakwest, the two filmmakers framed the issue around two immediate pressures, one artistic and one industrial: protecting emotion on screen, and managing a technical shift that is already touching writing, cinematography, sound, and postproduction.

WAiFF arrives as tools that can generate images, voices, textures, and animation—and streamline parts of production—move from novelty to workflow. The debate, as described at the festival, is no longer just about model performance. It’s about rules, traceability, ethics, and working conditions, as producers chase time and cost savings while creative and technical teams push for guarantees.

Claude Lelouch: Emotion can’t be “programmed”

Lelouch steered the discussion back to what he argues can’t be automated: the lived, accidental, emotional element that holds a film together. As reported, he didn’t dismiss the usefulness of new tools, but insisted on a hierarchy—technology should serve directing, not the other way around.

That stance reflects a worry shared by some directors: that AI could nudge filmmaking toward “optimal” images that are ultimately standardized, shaped by statistical taste rather than intention. In today’s use cases, the line can be thin. AI can suggest alternate shot breakdowns, simulate lighting moods, or help with virtual location scouting—assistive functions that resemble a sped-up storyboard.

For Lelouch, the problem grows when automation shifts from assistance to generation—when entire sequences are built without a clear anchor in a shoot, actor direction, or real set. He ties that anchor to believability: it affects whether an audience buys a scene.

He also framed the issue as a defense of crews. AI may promise shortcuts, but it moves work rather than erasing it. A “generated” shot still demands iteration, oversight, fixes, and a consistent style across a feature. The risk for productions, he suggested, is underestimating that workload and pushing it onto fewer people under tighter deadlines—leaving less time to search, fail, and try again.

In that view, AI can be useful for preparation, testing, previs, and technical cleanup—but it shouldn’t dictate the film. Responsibility still sits with people: if a scene moves an audience, it’s because a director, an actor, and an editor made choices and stand behind them.

Mathieu Kassovitz: AI changes who controls the pipeline—and who gets paid

Kassovitz approached AI as an industry force that can quickly rearrange tasks, budgets, and power. His central concern, as reported, was the production chain and the jobs inside it: who controls the tool, who owns the data, who signs off on results, and who gets paid for what.

He argued the most exposed roles aren’t limited to the obvious ones. Beyond visual effects, AI can be used in dubbing, cleaning audio tracks, restoration, compositing, and crowd animation. Writing is also implicated through tools that assist with synopses or dialogue.

Kassovitz emphasized a point he sees as widely misunderstood: AI isn’t just “software.” It can become a new organization of work, where creative labor fragments into micro-tasks—annotation, validation, correction.

He also raised the risk of normalized imitation. If a producer demands a shot generated “like” a specific reference, the tool can push toward copying, raising questions about stylistic diversity and the place of authors. Kassovitz argued for clear rules: contractually defining AI use, documenting what was generated, protecting moral rights, and setting limits on exploiting voices and faces.

His warning also touched on dependence. AI can lighten some roles, but it can also lock productions into platforms—often non-European—that set standards. In the short term, the temptation is speed. Over time, the question becomes strategic: what happens if prices change, access is restricted, or traceability requirements arrive too late? Kassovitz urged anticipating those risks to avoid irreversible choices made under pressure.

Rights, data, and traceability move to the center

Beyond artistic sensibilities, the WAiFF exchange pointed to a legal and economic fight: rights over works and over the materials used to train models. When images, scripts, music, or voices are used for training, traceability becomes decisive. Professionals want practical answers: can you prove what was used, under what license, and with what compensation for rights holders?

The issue is especially charged for voices and faces. An actor may accept traditional dubbing but refuse reuse of a synthesized version of their voice. A technician may accept an assistive tool but reject having a model “learn” from their creations without compensation. Festival discussions, as described, suggest contracts will need explicit clauses covering data use, duration, territories, and the possibility of withdrawal.

Without a framework, the risk is more disputes—or deals that disadvantage independent workers. Technical solutions exist in part, including watermarks, metadata, and fingerprinting systems meant to identify generated content. But they aren’t universal, and the chain of proof can break quickly when a sequence passes through multiple tools, vendors, and countries.

Economically, rights connect to value. If AI reduces certain costs, where does the margin go? In the best case, it could fund more shooting, rehearsals, sets, or high-end animation. In the worst, it becomes a tool to squeeze jobs and compress schedules. The festival conversation, as reported, stressed that technology doesn’t decide on its own—contracts, budgets, and producer choices determine whether AI strengthens creativity or weakens the ecosystem.

From storyboards to compositing, AI use is already spreading

The WAiFF debate is rooted in experiments already visible across production. Early on, AI is often used to generate storyboard directions, mood boards, costume variations, or set simulations—aimed at saving time, communicating intent, and reducing back-and-forth. For directors, that can mean testing options quickly before locking a plan.

During production and postproduction, uses broaden: image cleanup, stabilization, upscaling, denoising, audio source separation, and compositing assistance. Many of these tools existed in earlier forms, but generative models accelerate the pace and lower the barrier to entry. A small studio can attempt more ambitious effects; an editor can get a presentable cut faster for internal screenings.

But the convenience comes with organizational costs. Teams have to learn how to make requests, document decisions, and secure data. Confidentiality concerns rise fast when unreleased footage moves through cloud services. Industry participants have often stressed the need for local solutions—or, at minimum, strict protocols—especially because most incidents aren’t publicly discussed.

A major friction point remains audience perception. Some viewers accept AI for invisible fixes but resist it when it replaces an identifiable artistic gesture, like acting or animation. That reaction can shape distribution and marketing. WAiFF discussions suggest a growing push for transparency—some productions may eventually be expected, at least within the industry, to disclose where AI was used, similar to how certain visual effects or restorations are documented.

Frequently asked questions

What is WAiFF 2026, and why does it matter for film? The World AI Film Festival (WAiFF) 2026 brings together film and digital-creation professionals to debate concrete AI uses: which tools enter pipelines, what rules govern data and rights, and how jobs adapt to faster production practices.

Which film jobs are most affected by AI, according to the debate? The discussion highlighted writing (assistive tools), image postproduction (compositing, cleanup), sound (source separation, synthetic voices), animation, and some visual effects. The core issue is not only task removal, but shifting work toward validation, correction, and data management.

How can the industry limit risks around rights and data? The most cited paths involve clearer contracts (use, duration, territories), better traceability, and technical standards that identify what was generated—along with stronger security when footage or assets move through cloud platforms.

Can AI coexist with auteur filmmaking? Yes, if the tool serves a clearly assumed human intention—used for previs, testing options, speeding technical fixes, or improving team communication. Tensions rise when it imposes a standardized look or replaces creative choices without rules or transparency.

Key takeaways

At WAiFF 2026, Lelouch argued for protecting emotion and a connection to the real world against automation. Kassovitz focused on how AI could reshape jobs, contracts, and control of the production pipeline. Rights, data traceability, and the handling of voices and faces dominated concerns, as practical AI use expands from storyboards to compositing—bringing new security and standards challenges with it.

Sources

WAiFF 2026 : le regard de Lelouch et Kassovitz sur l’IA

Les rebonds de l’animation – A la Une – Mediakwest

CLAUDE LELOUCH AU WAIFF 2026

🎬 PRO TALKS — IA et création, et si on décidait enfin ? …

World AI Film Festival 2026 : quand l’IA fait son cinéma à …