How Award Juries Judge AI Work in 2026

Every major creative awards program rewrote its AI rules for 2026, and they rewrote them in the same direction. Cannes Lions, D&AD, Clio, and The One Show now require you to disclose generative AI at the point of entry, back your claims with a signed verification, and accept that undisclosed or misrepresented AI can get an award pulled after it is handed out. The trigger was a single scandal, and the response reset the entry process for everyone. Underneath the paperwork sits the more interesting question every jury is now answering out loud: they do not care whether AI was used, they care whether the thinking was worth it. This is what the rules actually require, how juries weigh craft against idea, and how to enter without stepping on a landmine.

TL;DR

• Disclosure is mandatory across Cannes, D&AD, Clio, and The One Show. Non-disclosure is grounds for disqualification or post-award withdrawal.
• The trigger was the 2025 DM9 scandal, where a Grand Prix was revoked after a case film was found to contain AI-manipulated footage simulating real news coverage.
• Juries are explicit that AI-versus-human is the wrong axis. They judge idea, insight, and craft. AI is a tool that either serves those or does not.
• Case films are now the highest-risk surface. The rules specifically target AI-edited or AI-faked case studies and false efficacy claims, not just the creative itself.
• Practical advice: keep an honest AI-use log from the first frame, get client sign-off on every metric, and never let AI touch the case film's factual claims.

The scandal that reset the rules

The 2026 tightening did not come from nowhere. At Cannes Lions 2025, the agency DM9 won the Creative Data Grand Prix for a Whirlpool/Consul campaign, then had it revoked after the festival concluded the case film contained AI-generated and manipulated content that simulated real-world news coverage and results. CNN Brasil discovered its broadcast footage had been altered and filed a complaint. DM9 ended up withdrawing three campaigns accounting for twelve awards in total.

The important detail: the problem was never that AI was used in the creative work. The problem was that AI was used to fake the evidence of the campaign's results. That distinction is the spine of every 2026 rule change. Juries are comfortable with AI in the making. They are not comfortable with AI in the proof.

What each program now requires

The four major programs converged, with local flavors.

Cannes Lions. Mandatory AI-usage disclosure during entry, with non-disclosure as grounds for disqualification or withdrawal at any stage, including post-award. A mandatory Code of Conduct, content-detection tools to flag manipulated entry materials, and a dedicated panel of AI and ethics experts to review questionable submissions. Human and AI fact-checking, stricter client sign-off, and bans of up to three years for agencies caught submitting faked work. The festival also published an Integrity Handbook and ran webinars through the spring.

D&AD. Structured disclosure of any AI involvement at the point of entry, across all categories. Crucially, the disclosure now extends beyond the creative output to all submission materials, including case films and supporting documentation. Every entry must carry an Entry Validation Card signed by a senior representative, functioning as a Code of Conduct declaration that all claims are accurate and verifiable.

Clio. Publicly "evaluating next steps" for its entry process and jury code of conduct in response to the Cannes fallout, moving in the same disclosure-and-verification direction.

The One Show. Tightening protocols to address AI-edited submissions and false efficacy claims, including adjusted client verification letters so that both client and agency must specifically affirm the accuracy of facts and metrics in the case study.

The common thread: the case film and its metrics are now the regulated surface, as much as or more than the creative. Agencies called the new requirements "burdensome but necessary." Both halves of that phrase are true.

How juries actually weigh craft versus idea

The rules govern honesty. The judging governs quality, and on quality the 2026 juries were remarkably consistent. A few on the record from Cannes:

Rakesh Menon: "Craft is not a layer you put on top; it has to be built into the idea itself," and "AI will always be a tool that can help enhance or even enable the idea," but "instinct is earned, not generated." Gurbaksh Singh: "The strongest work does not feel AI-made or human-made. It simply feels thoughtful, original, and relevant." Kainaz Karmakar framed the new AI craft subcategories as a chance to "create something awe-worthy."

Decode that and the judging rubric is clear. AI does not add points and does not subtract them. It is invisible to the score. What gets scored is whether the idea is worth the effort and whether the craft is built into it rather than sprayed on top. The proof is in what won: at Cannes Lions 2026, the most decorated AI-adjacent work was a human-crafted comedy campaign that argued against AI advertising. The AI Craft subcategories now sit inside the craft-led Lions, judged on the same terms as every other discipline, which is itself the statement.

The craft-vs-idea debate, honestly

There is a real tension the quotes paper over. If AI collapses the cost of execution, does craft still mean anything, or does everything now reduce to the idea? The jury answer is that craft moved rather than disappeared. It is no longer about who could afford the render. It is about the taste and intention applied to the render, the editing, the direction, the thousand small decisions a model does not make. When execution is cheap, craft becomes a judgment discipline instead of a budget one. That is harder to fake and easier to recognize, which is exactly why juries lean on it.

Practical entry advice

Five rules if you are entering AI-touched work in 2026.

1. Keep an AI-use log from frame one. Record which tools touched which assets, at what stage. You will need it for the disclosure field, and reconstructing it after the fact is where mistakes happen.
2. Never let AI near the case film's facts. This is the DM9 line. Generated or "recreated" footage that stands in for real results, real news coverage, or real reactions is the fastest way to lose an award and earn a ban. Recreate nothing that claims to be evidence.
3. Get explicit client sign-off on every metric. D&AD's Validation Card and The One Show's revised verification letters mean the client has to affirm the numbers too. Line that up before you submit, not after.
4. Disclose generously. Over-disclosure costs you nothing on the score, because juries do not penalize AI use. Under-disclosure is existential. The asymmetry is total, so err toward telling them everything.
5. Enter on the idea, not the technique. "We used AI" is not a case. The jurors are explicit that the tool is invisible to them. Lead with the insight and let the craft show. If the entry's whole pitch is the technology, it will not clear the bar.

What this means for brand teams

The awards rules are a leading indicator for where client and platform norms are heading. The same disclosure-and-verification discipline is settling into how you disclose AI in ads generally, and into the rights and IP conversations clients now open with. Building an honest AI-use log and a clean metric-verification trail is not just award hygiene. It is how you keep a campaign defensible when anyone, a jury, a client, a regulator, asks how it was made.

None of this is a reason to avoid AI in awarded work. It is a reason to be transparent about it and to make sure the idea carries. If you want a workspace where the whole generation process, every model, every version, every asset, is logged in one place so your disclosure writes itself, start on 8frame. For the campaigns clearing the 2026 bar, see the best AI-made ads of 2026 and what actually won at Cannes.