Since the beginning of September 2026, several European regulatory texts have come into effect, major platforms have modified their business models, and generative artificial intelligence continues to reshape digital marketing practices. The web landscape is changing rapidly, and some news deserves more attention than just a headline.
Mandatory labeling of AI content: what Article 50 of the AI Act changes concretely
The topic is not (yet) making headlines in mainstream media, but it concerns all digital professionals. Since August 2, 2026, Article 50 of the AI Act mandates machine-readable labeling of any content generated or significantly modified by artificial intelligence, whether it be text, image, sound, or video.
Public-facing chatbots must now inform the user that they are interacting with a machine. For systems already deployed before this date, a grace period runs until December 2, 2026: technical watermarking must be in place by then, otherwise sanctions will apply.
The penalty scale highlights the stakes: up to 15 million euros or 3% of global turnover for failing to meet transparency obligations. For a web agency, an online media outlet, or an e-commerce business that uses automatic product sheet generation, the issue is no longer technical; it is legal.
Those who wish to closely follow these regulatory developments and their impact on the sector can access the news page of Aura Blog, which regularly covers these topics.
Digital Omnibus Regulation on AI: a game-changing delay for B2B
Less publicized than the AI Act itself, the complementary regulation (EU) 2026/1744, published on July 24, 2026, has postponed the obligations applicable to high-risk AI systems. This text, sometimes referred to as the “Digital Omnibus on AI,” grants an additional deadline to B2B solution providers that integrate artificial intelligence into their digital services.
Specifically, SaaS software providers, marketing automation platforms, or predictive analytics tools now have more time to comply with documentation, data governance, and risk assessment requirements. Field feedback on this point varies: some publishers see it as a welcome respite, while others fear that this delay may encourage inaction.
The European regulatory framework around digital and AI has never been so dense. In just a few months, three major texts (AI Act, strengthened Digital Services Act, Digital Omnibus) overlap, each with its own implementation timeline.
Google AI Contribution and robots.txt: two signals for web publishers
On Google’s side, two announcements from September 2026 deserve to be read together. The first concerns the project called AI Contribution, a new payment model for publishers whose content feeds AI-generated responses. The operation relies on criteria that Google has not yet made public, raising legitimate questions about the transparency of the system.
The second news is more technical: a new setting allows publishers to publish a “no-training” directive in their robots.txt file. This directive blocks the use of content to train AI models while allowing Googlebot, Applebot, and Bingbot to crawl normally for indexing. In short, a site can now refuse to feed AI without disappearing from search results.
These two movements reflect a structural tension: Google needs publishers’ content to fuel its AI features, and publishers seek to maintain control over the value of their production. The available data does not allow for conclusions about the actual effectiveness of the AI Contribution program, but the mere existence of it signals a shift in the balance of power.
What publishers in France should watch
The combination of the AI Act and the changes to robots.txt creates a new legal and technical context. A French publisher who publishes content online must now balance three parameters:
- Allow or disallow AI crawling via robots.txt, knowing that blocking training does not block traditional indexing
- Check the compliance of their own uses of generative AI (product sheets, assisted articles, visuals) with the labeling obligations of Article 50
- Follow the Digital Omnibus timeline to anticipate documentation obligations if their tools integrate high-risk AI
September 2026 concentrates a series of regulatory and industrial decisions that, taken separately, seem technical. Put together, they outline a web where algorithmic transparency, data control, and content remuneration become the real substantive issues. Digital professionals who do not follow these developments risk discovering their consequences when they face them.
