Context Window 28
This edition covers the US House passing a tax bill with a ten-year moratorium on state-level AI regulation, OpenAI’s continued lobbying for Fair Use, Anthropic’s new Claude 4 hybrid models, the announcements from Google I/O including SynthID Detector, Shopify’s AI enhancements and MCP integration, Anil Dash on Model Context Protocol as a Web 2.0 moment, a new paper on AI in higher education, the Chicago Sun Times’s hallucinated book list, and Steven Bartlett’s fully AI-written-and-voiced podcast.
Yesterday, the US House narrowly passed President Trump’s sweeping tax bill, which includes a ten-year moratorium on individual states regulating AI systems. This echoes one of OpenAI’s key requests to the US AI Action Plan in March, which called for relief from “overly burdensome state laws”. It’s unclear how the administration will respond to another key ask: “Applying the fair use doctrine to AI is not only a matter of American competitiveness—it’s a matter of national security.” The direction of travel favours Big Tech, and every publisher should be considering what a Fair Use scenario looks like in terms of impact on licensing and stakeholders. Anthropic released its new Claude 4 hybrid models, which are capable of switching between quick and more reasoned responses. It’s more costly to use and may be overkill for everyday publishing tasks, but for more advanced coding and analytics tasks, its performance benchmarks look very strong. It’s been a big week for Google, which held its I/O developer event. Several new tools are promising for publishers creating interactive content, including new generative models like Imagen 4 (images) and Veo 3 (video with sound), enhanced audio and video summaries through NotebookLM, and creation of quizzes and infographics through the Canvas tool. To explore the announcements (and test one of the updated tools), Google has made the event data available through its NotebookLM research tool: you can ask questions of it here. Google also released a new tool, SynthID Detector, to help assess whether a particular piece of content was created with one of its AI tools. Google has previously acknowledged the limitations of detecting AI-generated text (you may have seen the meme this week of an AI detector concluding that the Declaration of Independence was written by AI). So this is not a deterministic tool, but a useful guide. Ecommerce leader Shopify announced a range of AI enhancements, including AI-assisted store creation—useful for publishers exploring direct-to-consumer channels—and integration with Anthropic’s Model Context Protocol to build AI agents such as book recommendation or customer support tools. Spinning up a store got a lot easier. On the subject of MCP, Anil Dash wrote an excellent explainer on the protocol and how it could support a more open, interoperable web. For publishers, the idea of resisting data centralisation and dominant platforms is both familiar and timely. The piece offers a more hopeful, open alternative to the consolidation implied in OpenAI’s lobbying. A new research paper reviews AI in higher education, highlighting potential for engagement and pedagogy, but calling for more empirical evidence. Academic and educational publishers may find opportunities here for developing AI-aligned materials—especially where human-AI collaboration enhances learning. More than two-thirds of books listed in a Chicago Sun Times feature were hallucinated by AI. The paper blamed a freelance supplier. It’s a cautionary tale for anyone outsourcing content creation—do your freelancer contracts address AI use explicitly? Entrepreneur and bestselling author Steven Bartlett released a new podcast written and voiced entirely by AI. One team member commented: “If you want to be an industry leader, you truly have to be pushing the boundaries of what’s possible in the industry and challenging every ‘norm’.” I suspect few traditional publishers would attempt something like this. The litmus test will be how the audience reacts. But it raises strategic questions about how traditional media can stay competitive as AI content evolves, especially if it leverages both brand and speed/low cost origination.
This was originally published in my email newsletter. To receive weekly updates on how AI is affecting the publishing industry, sign up here.