This week: AI can’t train for free anymore. We unpack Amazon’s licensing deal with The New York Times and what it signals for AI development , revisit how copyright law still applies in the age of generative models, and offer a hands-on prompt to test what tools your chatbot really has under the hood.

AI in Action

AI Training isn’t free anymore [Policy & Regulation]

In a quiet but consequential move, The New York Times has signed a multiyear deal with Amazon, allowing its journalism, recipes, and sports coverage to be used in Amazon’s AI products including Alexa, and to train its large language models. This marks the Times’ first official partnership involving generative AI, and it signals a strategic shift from fighting AI companies to working with them on its own terms.

The Times is still actively suing firms like OpenAI and Microsoft for unauthorized training. So why partner with Amazon? It comes down to control, compensation, and visibility. Under the agreement, Amazon’s models can surface NYT content in AI generated answers, but always with attribution and links back to the source. That’s traffic, transparency, and a revenue stream.

This is also a data quality move. As more AI models are trained on random internet scraps, The Times is offering vetted and trustworthy information, that benefits not just the publisher, but the user since licensed content tends to produce better, more accurate outputs. Amazon gets a stronger product. The Times gets paid. And the industry gets a blueprint for what ethical content partnerships can look like.

This could also mark a shift toward a tiered content economy, where top publishers monetize through licensing, while smaller creators may lack the visibility or leverage to do the same. As AI development becomes more regulated and reputation driven, more companies may follow Amazon’s lead, favoring licensing deals over legal risks. This deal signals that generative AI is heading toward attribution, licensing and paywalls for premium data.

Bits of Brilliance

The copyright debate [AI Concepts]

If news articles are available online, why can’t AI use them? It feels intuitive: if it’s free to access, it should be free to train on. However, when it comes to AI development, copyright law draws a much sharper line.

The rule is simple: access doesn’t equal ownership. You’re free to read a New York Times article online, quote it in a blog post, or share it with friends. But if an AI company uses thousands of those articles to train a model that later summarizes, paraphrases, or mimics the content? That’s classified as commercial reuse and it likely violates copyright unless permission has been granted.

It’s like playing a popular Coldplay song at your house party, no one’s coming to sue. But use that same song in a livestream or ad campaign? You’ll need a license. The same logic applies to written content. Just because it’s online doesn’t mean it’s free to mine for profit.

That’s why major publishers are striking deals. While The New York Times licensed its content to Amazon for Alexa and model training, it continues to sue other AI firms for unauthorized use. These licensing agreements aren’t just about legal protection, they also support better AI. High quality, licensed data can reduce hallucinations and misinformation, making outputs more accurate, grounded, and reliable.

As AI systems become trusted interfaces for news, research, and advice, the quality and legality of their training data matters more than ever. The next time you use an AI tool, ask: where is this answer coming from, and do the original creators know they’re part of it?

Curiosity in Clicks

What can your chatbot actually do? [Everyday AI]

You’ve probably heard that some AI chatbots can browse the web, analyze files, or even generate charts. Does your chatbot have those powers right now? Tool availability is the difference between chatting and doing. Knowing your AI’s capabilities up front lets you delegate smarter and avoid dead ends mid-task.

Prompt to try

“Summarize all the tools and capabilities you have access to in a markdown table. Include columns for: tool name, function, availability (✅ or ❌), and any limitations.”

Follow up with questions like: “What’s something useful you can do that most users overlook?” “What kinds of tasks are you especially good at with these tools combined?”

_{Byte-Sized Intelligence is a personal newsletter created for educational and informational purposes only. The content reflects the personal views of the author and does not represent the opinions of any employer or affiliated organization. This publication does not offer financial, investment, legal, or professional advice. Any references to tools, technologies, or companies are for illustrative purposes only and do not constitute endorsements. Readers should independently verify any information before acting on it. All AI-generated content or tool usage should be approached critically. Always apply human judgment and discretion when using or interpreting AI outputs.}