Meta's Money Muscle

AI Inside for Wednesday, July 2, 2025

Lots of interesting stuff this week, including Cloudflare’s new AI bot controls, a flurry of AI copyright lawsuits, and Mark Zuckerberg’s superintelligence obsession.

But first… a huge shoutout to our Patron of the Week, D J Rout! Support us at patreon.com/aiinsideshow.

Cloudflare’s New AI Crawler Marketplace

Cloudflare is launching a marketplace for publisher control over AI crawlers hitting its sites. New tools let users decide what kinds of AI bots can access their data, and monetize that access. By default, users can block AI bots from crawling their data. There’s a “pay-per-crawl” system for setting allowed actions and monetizing access, plus granular controls like whitelisting specific crawlers and different settings for training, fine-tuning, and inference. Upside: publishers get more control and potential compensation. Downside: parts of the web may close off, limiting research and the quality of AI models.

Creative Commons Launches CC Signals

Creative Commons has launched CC Signals, a new framework to guide how datasets can be used in AI development. The goal is to balance internet openness with the growing demand for data to train AI models. Data holders can specify if and how their content is reused by machines, using legal and technical tools similar to existing Creative Commons licenses. Public feedback is being sought ahead of an alpha launch in November 2025.

AI and Copyright: A Legal Tangle

#1: A group of authors sued Microsoft in New York federal court, alleging the company used nearly 200,000 pirated digital books without consent to train its Megatron AI model. They charge the AI with mimicking the style and content of those works and seek a court order to halt Microsoft’s actions, plus statutory damages of up to $150,000 per work.

#2: U.S. District Judge Vince Chabri dismissed a copyright infringement lawsuit from 13 authors, including Sarah Silverman, against Meta over using their works to train generative models. The judge said the authors “made the wrong arguments and failed to develop a record in support of the right one.” The ruling was narrowly applied to these plaintiffs and doesn’t mean Meta’s use was lawful. He also criticized Judge Alsup’s approach in the Anthropic case, saying judges are “fixated on personal bugbears rather than grappling with fundamental questions about how copyright should work in the age of AI.”

#3: Getty Images dropped its primary copyright infringement claims against Stability AI in the UK, mainly due to jurisdiction challenges. The case now focuses on whether importing and using AI models trained abroad constitutes infringement in the UK.

Senate Rejects Ban on State AI Regulations

The US Senate voted 99-1 against a 10-year ban on states enacting their own AI regulations, which was to be included in Trump’s major tax and spending bill. The bill passed, but this provision, supported by big tech, was stripped away. The Senate felt states should retain the ability to safeguard their residents regarding AI. Had it passed, states would have had to choose between enforcing their own laws or getting access to $500 million in federal AI infrastructure funding. But it’s all moot now, as the ban was removed!

Zuckerberg’s Superintelligence Obsession

Meta announced the Meta Superintelligence Lab (MSL), led by Alexandr Wang (ex-Scale AI) and Nat Friedman (ex-GitHub CEO). The move consolidates Meta’s AI teams to pursue superintelligence, with “hundreds of billions” invested in infrastructure, talent, and acquisitions. Meta is aggressively recruiting from OpenAI, Anthropic, and Google, reportedly offering up to $300 million over four years for top talent. Sam Altman responded on OpenAI’s internal Slack, calling Meta’s approach “somewhat distasteful,” adding, “Missionaries will beat mercenaries.” OpenAI is recalibrating its compensation to keep people in-house.

Microsoft’s Medical AI Outperforms Doctors

Microsoft is developing a new language model to diagnose disease, and CEO Mustafa Suleyman claims it’s four times more accurate than a panel of human physicians. Tested on 304 New England Journal of Medicine case studies, the model broke each case down step by step, then compared results with Gemini, ChatGPT, Claude, Llama, and Grok. The model achieved 80% accuracy, compared to 20% from a human panel, and picked far less expensive processes. Suleyman is all in on superintelligence: “This orchestration mechanism—multiple agents that work together in this chain-of-debate style—that’s what’s going to drive us closer to medical superintelligence.”

MIT Study: ChatGPT May Dull Critical Thinking

Fast Company’s Harry McCracken highlighted a new MIT Media Lab study. Participants wrote essays on SAT topics—some independently, some with search engines, others with ChatGPT. Brain activity was tracked. The study found that while ChatGPT makes writing easier, it may negatively impact long-term critical thinking. “By simplifying the process of obtaining answers, LLMs could decrease student motivation to perform independent research and generate solutions. Lack of mental stimulation could lead to a decrease in cognitive development and negatively impact memory.” McCracken reflected on his own AI use, ultimately seeing AI as a great starting point, not a solution tool.

The Velvet Sundown: AI Band or Not?

The Velvet Sundown, which sounds pretty indie rock, is by all accounts a band created by AI. All press photos appear AI-generated, and the music has a familiar AI sheen. Their Spotify listeners jumped from 325k to 634k in days. Their bio includes a Billboard quote that doesn’t seem to exist, and searches for band members come up empty. Deezer’s AI-detecting algorithms say the band’s music is AI-generated. The band claims, “we never use AI.” Their Twitter says, “Yes, We Are A Real Band & We Never Use AI #NeverAI.” They’ve locked personal accounts due to harassment. I’ve reached out to see if they’ll join us on the show. But really, if it IS AI generated and you like the music, what does it matter? I’m intrigued by a band differentiating itself by mimicking AI music.

If you’re enjoying the show, PLEASE leave us a review on Apple Podcasts! It’s a HUGE help and gets the word out!

Interview Tease: Rishad Tobaccowala on AI and Advertising

Last Friday, we recorded an interview with Rishad Tobaccowala, author of “Rethinking Work.” In this week’s episode, we play a clip from the full interview where Rishad talks about the use of tools like Veo 3 in redefining how the ad industry will operate in the near future. Expect the full interview to hit the feeds this weekend!

Executive producers of this show include:
DrDew, Jeffrey Marraccini, WPVM 103.7 in Asheville NC, Dante St James, Bono De Rick, Jason Neiffer, Jason Brady, and Anthony Downs!!