Blog

Four reflections on the EU AI Act through a strategic comms lens

The EU AI Act has come into force, and with it, a thousand LinkedIn posts speculating on its meaning. While the act itself focuses on fines, risk tiers, and legal liability, it is the reaction from the industry which is more interesting than the Act itself, notably Anthropic’s decision to roll out text watermarking. This is a move which goes well beyond compliance. Instead it looks like a statement of intent for setting a more responsible AI direction, actively embedding signals into AI-generated text as a mark of automated provenance.

It is somewhat lacking in nuance – this watermarking doesn’t care if you’ve entirely crafted your writing within an LLM, used it for editing support or sense checking. Still, this move takes things further than a response to fresh red tape. It marks a more fundamental shift in how trust is built, maintained, and communicated in an AI-saturated market. The decision issues a challenge to other LLMs, and places Claude streets ahead when it comes to answering the fundamental question lurking at the heart of AI – how far can I trust this?

The EU AI Act’s full application brings mandatory disclosures around potential deepfake imagery. There are exceptions for AI generated text when it has been subjected to human edits. Anthropic’s move to unilaterally digitally watermark has raised the bar further, and throws down the gauntlet for others to follow.

This is potentially a major inflection point for AI – a topic we have been monitoring closely as it has such important resonance for our industry and our clients’ reputations. Here are our initial thoughts on the AI Act, and what it means for your brand reputation.

1. The end of “AI hand-waving”: machine-readable transparency is integral to brand identity
Article 50 of the Act mandates clear labeling for synthetic media, deepfakes, and automated interaction systems. This legal requirement is straightforward: tell people when they are interacting with or consuming content generated by Artificial Intelligence.

The strategic opportunity lies in how you act on that requirement.

Treating disclosure as a fine-print legal disclaimer signals anxiety about your offering. It reads as hiding AI use away, buried in the small print. Instead, integrating transparency directly into the brand experience radiates confidence. Everyone is using AI. It’s not a de facto signal of laziness from a brand. But clarity of use will go a long way. Treat mandatory disclosures as a badge of quality and authenticity, and progressive brands can turn regulatory compliance into a clear proof point of corporate integrity.

Our advice? Audit external-facing touchpoints today. Map out how automated disclaimers can be styled and communicated to reinforce brand trust, rather than undermine it.

2. Is the medium the message? What is the watermark saying?
Anthropic treats all AI fingerprints on a piece of content as the same. Whether that’s assessing how well a draft stacks up in terms of narrative, sense checking an award entry against entry criteria, or full production from scratch – those flags will be embedded in the end result text, for those who care to look for it.

It’s how people respond to this move that matters. Will checking for AI production become more operationally embedded, or has its use already become so ubiquitous that it’s assumed? For awards entries, verifying that case studies are accurate has already become key – yet at the same time some awards bodies have already integrated their own AI engines into the submission process. Does it matter if the words have been polished or fully produced by automation if they move you?

All of these questions come down to very individual feelings when you arrive at a piece of content, and what you expect from it. If we’re talking about text for a website whose primary function is to simply explain complex products and services to an audience and to connect to wider web functionality at the same time, flags for AI probably matter less. The truth behind those words matters most. Claiming what is not the case is never acceptable, and checking carefully for hallucinations should remain an important human  proofing task.

3. The first-mover advantage: Turning friction into market position
Regulatory changes always create market noise and executive hesitancy. The instinct among many corporate leaders is to wait, observe how peers react, and adjust at the last possible moment.

That delay represents a major strategic misstep.

Get ahead of the pack and show you have understood the assignment. When enterprise RFPs begin demanding clear EU AI Act documentation, compliant businesses will progress rapidly while competitors remain stuck in legal reviews.

Seize the opportunity. Position your brand’s regulatory readiness as a core pillar of your market differentiation in upcoming PR campaigns, investor decks, and industry keynotes.

4. Think beyond borders

Considering this as ‘just’ EU legislation is missing the point. Just because your region isn’t under its purview ignores the emerging bigger picture. Claude has proven that. This is the first major international statement of intent about the responsible deployment of AI.

Smart strategies will see it as the windsock indicating strength of sentiment, and base foundation thinking for company strategies accordingly. Embed the trust now with open statements of AI use, and customers will respond. We’ve seen this ourselves with Stephen Jenkins’ LinkedIn disclosures of AI assistance – framed in his own words, and recognising that just because everyone is doing it, doesn’t mean the implementation is always right for us.

Borders exist for the business too. Be your authentic brand self, and open up. Do it now. The assumption is that everyone is already using AI. Your words will count for more when they are demonstrably your own. People will respond, and connect accordingly.