If you've logged on to LinkedIn over the last few days, you likely already know the headlines from Cannes even if you haven't opened any recaps:
AI dominated nearly every panel, beach, and product demo for at least the third year running;
Creators arrived in force, with some 500 of them (and especially in the evening party queues, it felt like more) descending on the Croisette; and
"Craft" was framed as the antidote to machine-made monotony
I’ll draw on these themes too, but will try to expand a bit beyond “here’s what you missed” (as I’ve never been particularly concise).
What stood out to me compared to past years was the change in tone: the AI conversation evolved from "look what it can do" to "has any of it worked, and can we trust it?" While the creator presence remains a testament to the growth and importance of the creator economy, many discussions this year centered on creators’ ability to bring authenticity to brand messaging specifically in the face of AI.
I’ll leave the “has any of it worked?” discourse to others, and focus below on covering two sides of the AI trust conversation, along with how authenticity was framed as a differentiator in an increasingly synthetic world.
To the extent that Cannes has a safety and integrity track, this year it focused heavily on AI transparency along two dimensions: consumer-facing transparency through provenance labels, and advertiser-facing transparency through measurement (or the lack thereof).
The reason for the former is no mystery. A wave of regulation requiring AI disclosure labels is on the horizon. The European Union (EU) AI Act will require the labeling of so-called deepfakes from August, and California's AI Transparency Act takes effect the same day. New York's synthetic performer law took effect in June, mandating disclosure for AI-generated performers in ads that run in the state, though notably without specifying what disclosure should actually look like. And that’s the crux of the problem. Whether they apply to AI providers or advertisers, these laws offer little guidance on how to apply their requirements in practice, which leaves a great deal open to interpretation.
The result is a fragmented compliance landscape.
In the absence of legal clarity, many marketers are looking to voluntary guidance such as the IAB's AI Transparency and Disclosure Framework, which calls for disclosure when AI materially changes identity or authenticity. Research from the WFA published in April shows similar areas of agreement around disclosure, finding that 96% of brands think an AI voice that could pass for human should be disclosed and 91% say the same of a synthetic human in a leading role, while just 4% think an AI-generated background needs one. There’s more debate around disclosing AI usage for other elements, like product images, but that’s not stopping brands from leaning into AI in their creative workflows anyway: 78% are using AI in consumer-facing work, according to the WFA, even as eight in ten say they still want clearer disclosure guidance.
The platforms are navigating this ambiguity as well, defining their own approaches and effectively acting as de facto regulators in lieu of greater specificity from lawmakers. There has been meaningful collaboration in places, most visibly on shared content provenance standards like Coalition for Content Provenance and Authenticity (C2PA), but much of the work still happens in silos. The tension that has always existed between competitors collaborating on shared problems is sharpened by the all-out race to “win” in AI.
My sense from the week is that this patchwork is here to stay for a while, and that the decisive question will be whether and how governments choose to enforce these laws, rather than whether everyone can get their act together ahead of time.
The second half of the AI transparency conversation centered on advertiser-facing measurement in AI environments. Apart from AI usage in ad creative, questions also emerged around how advertisers will measure and verify the ads they run inside AI environments. This applies across a wide range of measurement considerations, but for me (and for BSI), the question is particularly pointed where it touches brand safety and suitability in fully AI-generated environments such as chatbots.
What does "brand safety" mean when there’s no fixed page to evaluate? In a conversation, there is no URL to block and no static context to assess; the content that sits beside an ad is generated dynamically, and a single thread can swing from harmless to sensitive in the span of a few messages. What happens to ads that were previously rendered if the conversation takes a turn? The question of what "adjacency" means in that setting, and what, if any, reporting an advertiser receives on placements in private conversations, remains an open one.
Beyond adjacency, the private, conversational nature of these platforms raises deeper questions. The ways in which people have come to use chatbots give platforms an unprecedented window into what people want, moment to moment, which creates a remarkable trove of consumer intent for advertisers. OpenAI was on the ground this year and highlighted its position at the “intent” stage of the funnel. What remains unclear is how much of that trove advertisers will be allowed to target against, and, even where they can, how much transparency they can expect in return — which signals were used to place their ads, and what kinds of conversations they ultimately appeared inside.
The companies leading here have said many of the right things about keeping ads out of the organic response and steering clear of sensitive contexts, but public-facing policy pages are a long way from measurement an advertiser can independently verify, and right now advertisers have almost no visibility into the environments their ads run in. OpenAI made clear that advertising is a model ChatGPT intends to build and grow. As they and other AI platforms expand their foothold in the advertising space, the industry will have to reckon with many of the same measurement problems we faced with walled gardens and the open web, now in a setting that is even more opaque, and that demands exactly the kind of cross-industry collaboration the AI arms race discourages.
Both halves of the AI conversation come back to trust, and how difficult trust has become to establish in a world flooded by AI. But just as visible as the desire for transparency into where and how AI has been used is consumer distaste for any AI usage at all, transparent or otherwise. A 2023 IPA survey found that 74% of consumers said brands should be transparent about their use of AI. Yet even transparent AI usage can still work against the brand. New research unveiled at Cannes by The Harris Poll, the 4As, and Infillion found that 78% of consumers say AI makes ads feel less authentic, 73% are less likely to trust an ad they believe was made with AI, and 63% are less likely to buy from a brand that uses AI-generated ads. Even if brands deliver the transparency consumers say they expect for AI usage, AI usage itself still poses a risk.
For some brands, the solution is opting out of synthetic creative altogether: a handful of advertisers, Aerie and Le Creuset among them, have begun running "no AI" pledges as a selling point. Others are leaning on creators. The growing creator presence at Cannes is not new, but the way brands think about them has shifted meaningfully. Influencer marketing was once treated as a tactic to bolt onto a campaign; today it is increasingly a core channel in its own right, with U.S. creator ad spend projected to approach $44 billion this year, according to the IAB.
The logic has moved from reach to relationships; where brands once partnered with creators for the size of their audience, they now do so for the creator's standing with that audience, and the trust it represents.
The same lessons on AI usage that apply to advertisers apply to creators too. A study published this spring in the Journal of Consumer Research, drawing on eight experiments and more than a million TikTok posts, found that when audiences know a post was made with AI, they engage with it less, giving it 7-8% fewer likes, and they perceive the creator as having put in less effort and feel less connected to them. These findings highlight the premium consumers place on visible, human effort for advertisers and creators alike.
That, in my opinion, is why the week felt so heavily creator focused: it’s the flip side of the AI wave. The more capable the industry becomes at mass-producing synthetic content, the more it needs human creators to connect with audiences wary of anything that feels machine-made.
In this recap a year ago, I wondered whether an overreliance on synthetic creative might one day become a reputational risk in its own right. A year on, it increasingly looks like it can. As AI-generated content floods our feeds, our ads, and now our private conversations, the demand for authenticity, transparency, and humanity is rising to meet it with equal and opposite force. Transparency into AI usage is important and expected, but it’s not enough. The more synthetic the online world becomes, the more people crave humanity.
I’m sure we’ll still be discussing this tension on the Croisette a year from now.