1. The paradox nobody in tech wants to name

There is a sentence in the Alliance de la Presse's Perspectives #7 that deserves to be framed and hung on a wall. Major AI players keep arguing publicly against any regulation of copyright, while those same players, quietly, keep signing licensing deals with publishers. OpenAI with Le Monde in spring 2024. Meta with the Figaro Group in March 2026, a three-year contract covering archives and fresh content. Qwant running a nine-month pilot with twenty Alliance publishers.

Why the split? Because these companies know something their communications departments work hard to hide: a language model without reliable editorial content is a fast track to confident nonsense. The quality of answers depends on the quality of training corpora and the quality of real-time retrieval sources. And that quality, today, is produced by newsrooms.

Meanwhile, on the traffic side, the asymmetry is almost comical. According to the Tollbit report cited by the Alliance, chatbots generate 0.12 percent of publisher traffic. Not twelve percent. Zero point one two. Your content gets consumed at scale, repackaged to an audience that never comes back, and somehow you are told you should feel grateful to be contributing to innovation.

2. Three laws, three jurisdictions, one direction

What makes the Alliance's brief worth reading is that it puts three initiatives back into the same frame. Taken separately, they look like isolated wins. Taken together, they describe a regime.

The Voss Report at the European Parliament

Adopted on March 10, 2026 with 460 votes in favor and 71 against, the Voss Report states that companies developing AI systems must obtain licenses and pay the creators whose work feeds their models. The important detail is the shift in burden of proof: it is no longer up to the publisher to prove their content was used. It is up to the AI company to prove it respected the law.

The UK Creative Content Exchange

After a public consultation that received tens of thousands of responses, with 88 percent of respondents in favor of mandatory licensing, the UK government dropped its plan for a broad copyright exception for AI training. Instead, a Creative Content Exchange will run as a pilot platform to organize access to content in exchange for fair remuneration. The signal is strong: even in a country generally friendly to the tech ecosystem, public opinion has made up its mind.

The Darcos-Balanant bills in France

This might be the most interesting piece legally. Passed unanimously by the National Assembly (the Balanant side, on enforcement of neighbouring rights) and then the Senate (the Darcos side, on presumption of AI use), the combined framework introduces a simple and formidable mechanism. If a publisher produces a body of indicia suggesting that their content was used to train or feed an AI, it is up to the AI company to prove otherwise. France's audiovisual regulator, Arcom, gets an arbitration role with the power to issue financial sanctions.

Three texts, three methods, one philosophy: recognize that journalistic content has value, organize its payment, and give publishers real legal tools that no longer depend on their ability to fund years of litigation.

For deeper dives, we covered the Balanant vote in our March 29 analysis, and the UK framework in our breakdown of the CMA investigation.

3. What the signed deals already tell us

A question that keeps coming up in newsrooms: "What is a deal with an AI company actually worth?" The Alliance doesn't publish numbers (they are confidential), but the list of signed deals is information in itself.

• OpenAI / Le Monde: multi-year deal signed in spring 2024, renewed since. The first major agreement between a French publisher and a top-tier generative AI player.
• Meta / Groupe Figaro: three-year contract signed in March 2026, covering both archives and the ongoing article feed. A meaningful signal, since Meta had refused to engage under the neighbouring rights framework.
• Qwant / 20 Alliance publishers: a nine-month pilot to measure the actual value of journalistic content inside an AI-enhanced search engine. Less covered in the press, more interesting in the medium term.

Every deal sets a precedent. Every precedent anchors a value. And the more values get anchored, the harder it becomes to argue that editorial content is worthless. That is exactly the movement the Alliance wants to accelerate with the collective mandate it secured in November 2025 from 88 publishers.

4. The neighbouring rights lesson: a law without leverage is just paper

This is the part of Perspectives #7 that should be read on repeat in every newsroom. The Alliance reminds us that France transposed the European directive on neighbouring rights in 2019. On paper, a victory. On the ground?

• Google had to be hit with a 500 million euro fine before it took negotiation seriously.
• Microsoft has been stretching the conflict for seven years. In February 2026, the Paris court labeled its strategy "manifestly dilatory".
• Meta cut its first proposal by a factor of twenty compared to the initial figure.
• LinkedIn and X still flatly refuse to negotiate.

The law does not write the check. What writes the check is the combination of a legal framework, a way to prove usage, credible collective bargaining, and political will to actually swing when it matters. One without the others never works.

For publishers, the takeaway is very concrete. In 2026 as in 2019, the real work starts after the vote. And the first brick of that work is evidence. No evidence, no leverage. No leverage, no negotiation. No negotiation, no revenue.

5. Where Senthor fits in

At Senthor, we built the platform around a simple observation: regulation is moving in the right direction, but it moves slowly, and no publisher can afford to wait three years for a court ruling before starting to monetize. You need to act now, with the tools that exist now.

Concretely, here is what Senthor brings to a newsroom or a press group reading Perspectives #7 and wondering where to begin.

Detect, identify and qualify AI traffic

Senthor analyses the traffic hitting your content in real time and identifies training bots, AI indexing crawlers, and queries coming from conversational agents. No vague promises: dated, sourced, exportable logs. Exactly the kind of data the Darcos bill calls a "body of indicia". You move from an intuition ("I think ChatGPT is scraping my articles") to a dataset you can actually use.

Monetize without waiting for the laws

Senthor is already connected to the x402 protocol and Stripe MPP. Translated for a publisher: when an AI agent comes to fetch content on your site, you can answer "pay first". It pays in stablecoin. You receive euros on your bank account. No crypto wallet, no legal friction, no deal-by-deal negotiation.

Negotiate from facts, not impressions

Whether you go the contractual route (a direct deal with an AI company, like Le Monde or Figaro) or the regulatory route (an action before Arcom, a claim under the Darcos framework), you will need the same ingredient: proof. Senthor reports are designed to slot into an arbitration file, a commercial negotiation, or a collective action led by a body like the Alliance.

6. What to do now, concretely

If you run a newsroom, a press group or an independent outlet, here is the roadmap we suggest, independently of Senthor, after reading Perspectives #7.

1. Map your exposure. Which AI bots are crawling you? How often? On which content? As long as you don't know, you can't negotiate anything.
2. Build an evidence capability. The presumption of use is a powerful legal tool, but it starts with a body of indicia. You have to be able to produce those indicia.
3. Open a direct monetization channel. In parallel with collective negotiations, test at least one agent-to-agent payment path. Volumes are small today. The curves climb fast.
4. Don't leave the negotiation to others. Individual deals struck by big groups set the standards for the entire market. If you are not at the table, you accept what gets decided there.

Perspectives #7 ends on a cautious but lucid note. The laws are here. The tools exist. The moment is good. What remains is not to miss it, the way we almost missed neighbouring rights back in 2019.

To read the Alliance de la Presse original brief, head to their Perspectives newsletter. We recommend reading it in full.