How 'Really Simple Licensing' Is Rewiring the Web to Charge AI for Content Scraping

For decades, the internet operated on an unspoken agreement: search engines crawled websites for free, and in exchange, they sent traffic back to the publishers. But the rise of generative artificial intelligence broke that pact. AI models scrape content not to direct users to a source, but to answer questions directly, keeping users on their own platforms.

In response, publishers began locking their doors using the decades-old robots.txt file, a blunt instrument that simply tells bots to stay away. Now, a new standard called Really Simple Licensing (RSL) is rewiring the web's infrastructure, shifting the paradigm from a binary 'keep out' to a nuanced 'pay to enter.'[1]

The traditional robots.txt protocol was designed in the 1990s as a voluntary honor system. It allows a webmaster to list which parts of a site a crawler can or cannot visit. However, it offers no mechanism for monetization, and as AI companies grew desperate for training data, an estimated forty percent of stealth AI bots began ignoring these directives entirely.

Enter RSL, co-created by Eckart Walther, one of the original architects of RSS. Instead of a simple block, RSL uses an XML-based document format that allows publishers to attach machine-readable licensing terms directly to their content.[2]

Under the RSL 1.0 specification, a publisher can drop a license file onto their server or embed tags directly into their HTML headers. These tags dictate the exact terms of engagement for an AI crawler.

A news outlet might specify that their articles are available for a 'pay-per-crawl' fee of a fraction of a cent, or a 'pay-per-inference' royalty that triggers whenever an AI model uses their data to generate an answer. Major platforms including Reddit, Yahoo, and Medium were among the first to adopt the standard, signaling a massive shift in how digital property is treated.[4]

When analyzing the trade-offs of these two systems, the case for the traditional robots.txt approach rests on absolute simplicity and universal legacy support. For this older method, the primary advantage is that every web crawler on earth understands a basic disallow command, and implementing it requires zero technical overhead or legal negotiation.

The evidence supporting this approach is its thirty-year track record of keeping honest search engines in check. However, the argument against robots.txt is that it leaves money on the table and is entirely toothless against rogue scrapers. It forces publishers into an all-or-nothing defensive posture, where they must either give their data away for free or block the very AI systems that are becoming the internet's new front door.[3]

Conversely, the case for Really Simple Licensing centers on creator empowerment and revenue generation. For publishers, RSL transforms passive content into an active, monetizable asset. The evidence for its efficacy is already visible in the rapid formation of the RSL Collective, a clearinghouse that acts much like ASCAP or BMI in the music industry, pooling the bargaining power of thousands of independent creators to negotiate bulk payouts from AI giants.

The argument against RSL, however, is the friction of enforcement. Because RSL is still a signaling tool, it relies on AI companies actually agreeing to pay the tolls. If an AI developer chooses to ignore the XML file, publishers still have to rely on server-level IP blocking or prolonged litigation to enforce their terms.[1][2]

The financial implications of this shift are staggering. Industry analysts note that 2026 is bringing a hard reset to AI licensing, driven largely by the legal risks of unauthorized scraping and new regulatory pressures like the European Union's AI Act. By adopting RSL, publishers are establishing a clear, auditable paper trail of their terms.

If an AI company scrapes an RSL-protected site without paying, the publisher has concrete proof of willful infringement, dramatically increasing their leverage in court. This dynamic is forcing AI developers to the negotiating table, shifting the internet from an open buffet to a metered utility.

Ultimately, choosing between these protocols depends entirely on a publisher's goals. The traditional robots.txt block fits well when a website contains highly sensitive personal data, when the creator is philosophically opposed to AI training under any circumstances, or when a small hobbyist site lacks the infrastructure to manage micro-transactions. In these scenarios, a hard block paired with aggressive firewall rules remains the most straightforward defense. It provides a clear, unambiguous boundary that requires no ongoing management or legal oversight.[3]

On the other hand, Really Simple Licensing fits well when a publisher produces high-value, proprietary information—such as journalism, technical documentation, or specialized datasets—and wants to participate in the AI economy rather than just fighting it.

It is the ideal solution for media companies, independent writers, and data platforms that recognize AI as the next iteration of search and want to ensure they are compensated for fueling it. As the web transitions into an AI-first ecosystem, RSL offers the first scalable bridge between the companies building the models and the creators supplying the knowledge.[1][4]