How to measure the gap between the declared adoption of digital tools and their actual integration into business processes? Between Google’s announcements about compensating publishers through generative AI and the reality of usage in companies, the web news of 2026 reveals very contrasting dynamics depending on the size of the organizations and the type of technologies deployed.
Generative AI in business: declared adoption vs. actual integration
The figure is circulating everywhere: more than two-thirds of French companies with at least 20 employees report using generative AI at the beginning of 2026, according to the Banque de France. This data masks a structural gap.
The Banque de France specifies that usage remains largely confined to simple cases: writing, summarizing, document preparation. The functions concerned are mainly commercial, administrative, and financial, far from core business processes.
On the side of organized integration into information systems, Insee data shows a progression from 6% in 2023 to 18% in 2025 for AI technologies. The jump seems rapid, but it concerns less than one in five companies. The gap between “testing ChatGPT at the office” and “integrating a model into a production line” remains considerable.
Several specialized monitoring services document this digital news, and the Geek Wise website notably aggregates the technical developments of the web and digital tools that accompany these transformations.
| Indicator | 2023 Value | 2025-2026 Value | Source |
|---|---|---|---|
| Companies (20+ employees) reporting using generative AI | Not documented | More than two-thirds | Banque de France |
| Organized integration of AI in information systems | 6% | 18% | Insee |
| Main usage functions | – | Support, sales, admin, finance | Banque de France |
This table summarizes the central gap: massive surface adoption coexists with a slow transformation of internal workflows.
Robots.txt and no-training directive: what the new technical landscape changes
A recent adjustment allows publishers to publish a “no-training” directive in their robots.txt file. The principle: prohibit the use of their content to train AI models while allowing Googlebot, Applebot, and Bingbot to continue crawling for traditional search.
This separation between indexing crawl and AI training crawl creates a new layer of technical management for webmasters. Until now, blocking a robot meant also losing visibility in search. The no-training directive dissociates indexing and training for the first time.
At the same time, Google is testing a program called AI Contribution, a compensation model for publishers for generative AI. The criteria remain opaque, raising questions about the transparency of the economic model proposed to content creators.
Consequences for website publishers
For sites that produce original content, these two developments change the strategic arbitration. Accepting training crawl could open access to compensation via AI Contribution. Refusing it via no-training protects the content but closes that door.
No public scale has been communicated by Google regarding the amounts at stake. The opacity of the compensation criteria remains the main friction point for publishers trying to assess the program’s interest.
Digital tools 2026: website, email, and short video against ROI
The website, blog, and SEO continue to be regularly cited among the marketing channels with the best declared ROI, ahead of paid social and organic social. This ranking has not changed for several years.
- Email marketing maintains a high ROI, but production is accelerating due to AI, with increased risks of subscriber fatigue and deliverability issues
- Short video (Instagram, TikTok, YouTube Shorts) is becoming a central discovery format for brands, with a dominant presence in declared marketing usage
On the other hand, multiplying platforms without a measurable strategy remains the main trap. The challenge for companies is not to be present everywhere but to concentrate resources on channels with demonstrable returns.
Cybersecurity and regulation: the framework that structures digital news
The European regulation on AI (AI Act) is gradually coming into effect. France is simultaneously strengthening its regulation with measures targeting the security and accountability of artificial intelligence systems deployed on the market.
For companies, compliance is becoming a full-time job. Obligations vary according to the risk level of the AI systems used, which requires a prior audit of the digital tools in place.
Digital transformation of micro and small businesses
The gap is widening between large companies and small and medium-sized enterprises in the adoption of AI. The former have dedicated teams for integration and compliance. The latter often experiment without a structured framework, exposing them more to regulatory risks.
Available data shows that real digital transformation is progressing much more slowly than the discourse on innovation. Between the announcements of new tools and their actual deployment in business processes, the time lag is measured in years, not months. Companies that derive measurable benefits from these technologies are those that have invested in training and organized integration before piling up software subscriptions.
