A new licensing standard called Really Simple Licensing, or RSL, aims to empower web publishers by allowing them to set terms for how artificial intelligence developers can use their content. Major brands including Reddit, Yahoo, Medium, and Quora have announced their support for this initiative, which builds upon existing web protocols to enable publishers to specify licensing and royalty terms for AI data scraping. The RSL Standard allows site owners to charge fees for AI bots crawling their sites, either through subscription models or pay-per-crawl fees. This initiative comes amid ongoing legal disputes between AI firms and content creators, highlighting the need for a unified approach to content licensing. According to the RSL Collective, which is spearheading this movement, their goal is to create a scalable business model that facilitates fair compensation for content creators, ultimately addressing the legal gray areas surrounding AI training data usage.
Microsoft is taking a significant step to reduce its reliance on OpenAI by partnering with the startup Anthropic to enhance its Office 365 applications. The tech giant plans to incorporate Anthropic’s advanced models for features in Word, Excel, and PowerPoint, following months of negotiations with OpenAI regarding its restructuring plans. According to sources involved in the initiative, Anthropic’s technology has outperformed OpenAI’s in automating tasks like financial functions in Excel and generating visually appealing PowerPoint presentations. With over 100 million customers using Office 365 Copilot, Microsoft aims to maintain the current subscription price of $30 per user per month while integrating these new capabilities.
The United Arab Emirates has unveiled a new open-source artificial intelligence model called K2 Think, which is designed for advanced reasoning and rivals the capabilities of leading models from the United States and China. Developed by researchers at the Mohamed bin Zayed University of Artificial Intelligence, this model has 32 billion parameters and demonstrates significant technical innovations, including effective simulated reasoning and agentic planning processes. Eric Xing, the university’s president, emphasized that K2 Think’s ability to perform complex reasoning tasks is comparable to larger models with over 200 billion parameters, such as those from OpenAI. The model’s development involved thousands of graphics processing units and was optimized for efficiency on specialized Cerebras hardware.
Alibaba’s Qwen Team has launched its latest large language model, Qwen3-Max-Preview, boasting over 1 trillion parameters, making it the company’s largest model to date. This new model outperforms its predecessor and competes closely with leading models in the industry, showcasing strong performance across various benchmarks. The model features a maximum input of 258,048 tokens and a maximum output of 32,768 tokens, significantly enhancing its context-handling capabilities. Initial tests indicate that Qwen3-Max-Preview not only responds faster than competitors but also avoids common pitfalls seen in language models. However, it is important to note that while the model is available through various platforms, it has not been released under an open-source license, requiring users to access it via paid API options.
Why do we care?
So now we’ve got a “Really Simple Licensing” standard—Reddit, Yahoo, and others want to make AI companies pay when scraping their sites. Sounds great, but unless there’s teeth, big AI firms may just ignore it. The content side will need to band together here.
And this isn’t just about big publishers. Plenty of SMBs—think local news sites, niche blogs, training firms, even MSPs producing thought leadership—are generating content that AI companies quietly ingest. If RSL gains traction, those smaller players could finally have a mechanism to say, “our work has value,” and either control usage or demand compensation. For service providers, the play is two-fold: protecting your own IP and advising customers who publish content on how to leverage or defend against these new licensing rules.
Microsoft, on the other hand, isn’t waiting around. They’re sliding Anthropic into Office 365, because hey—maybe Claude does PowerPoint better than GPT. That’s real, and it’s in front of customers.
And sure, the UAE and Alibaba are waving around new big models, but unless your SMB customer is running sovereign AI or wants a trillion-parameter Chinese chatbot, that’s not today’s problem. What matters is this: Microsoft’s AI keeps changing under the hood, and RSL might change the rules of what content AI can touch. Those are the moves worth tracking.