AI Inside: Meta’s AI Ad Takeover

AI Inside for Wednesday, June 4, 2025

Lots of interesting stuff this week, including Meta’s push to fully automate digital ads with AI, the New York Times finally signing a generative AI deal with Amazon, and major music labels negotiating with AI music startups.

But first… a big shoutout to our Patron of the Week: Steve Remington! If you want to support the show and get cool perks, join us at patreon.com/aiinsideshow.

Meta’s AI Ad Revolution

Meta is moving to fully automate the creation and targeting of digital ads using AI by the end of 2026. Right now, Meta’s tools can recreate new variations of existing ads, but the next phase is letting brands upload a product image and budget, then letting AI generate all aspects of the ad, visuals, video, text, target audience, even budget allocations. Ads will be personalized in real time, so different users will get different ads based on immediate factors like location. With advertising making up more than 97% of Meta’s revenue, this move brings AI even more central to its business.

Meta’s Internal Risk Assessment Goes AI

Meta is also shifting its internal risk assessment process toward AI automation. Up to 90% of its decisions about privacy, societal risk, and content integrity will be AI-driven. Algorithm changes, new safety features, and content sharing rules will soon largely be approved by AI systems, including areas like youth safety, AI safety, and violent content. Meta says human experts will still handle novel or high-risk issues, but this raises questions about whether this will reduce scrutiny and make it less likely that negative outcomes are caught in advance.

The New York Times Signs AI Deal with Amazon

The New York Times, one of the most notable holdouts in AI licensing deals, has signed its first generative AI deal with Amazon. Amazon gets access to Times content including news articles, NYT Cooking recipes, and The Athletic to power Alexa and train its AI models. It’s a multi-year agreement, reflecting a broad and growing trend of publishers striking licensing agreements with tech firms. No financials disclosed. Placing value on its content in a deal like this also potentially influences its legal battles, as NYT is still suing OpenAI and Microsoft for alleged copyright infringement.

Music Labels in Talks with Udio and Suno

Universal Music Group, Warner Music Group, and Sony Music Entertainment are in advanced talks with Udio and Suno, potentially licensing their music catalogs for training and use. Labels are reportedly pushing for license fees, a small equity stake in each company, and greater control over how their music is used within those AI products. Possible implementation of fingerprinting and attribution systems to track and compensate artists. Suno and Udio want flexibility to experiment and financial terms that work for startups. Finding terms that make everyone happy. That’s the real challenge here.

Google’s AI Edge Gallery Quietly Launches

Google quietly launched a new app called AI Edge Gallery, meant for running on-device AI models directly on the phone. Android now, iOS coming soon. It’s entirely offline, and open source models from platforms like Hugging Face can be installed and used for image analysis, interactive chat, text summarization, and code generation. Requires a premium device with a fast CPU/GPU and at least 6GB RAM. It’s experimental and only available by sideloading the APK via GitHub along with a Hugging Face account.

Mary Meeker’s Massive AI Trends Report

Mary Meeker is a bit of a tech prophet when it comes to trends. She dropped her first Trends report in four years and this time around, has 340 pages worth of analysis on AI’s global impact in 2025. There is a ton to pour through in this report. Jeff and I danced through a very small percentage of it on the podcast.

Chefs Using ChatGPT for Culinary Creativity

ChatGPT is becoming a collaborator in master chef’s kitchens. The New York Times covers how big name chefs are experimenting with generative AI for inspiration (brainstorming recipes), streamlining operations (improving the technical process of sausage making, for example), enhancing menu development, and even restaurant design. One chef, Grant Achatz from Chicago’s Next restaurant, is creating fictional chef avatars and prompting ChatGPT for the kinds of recipes that fictional chef might make, resulting in a nearly AI-composed 9-course menu. In most cases, AI is used as a creative assistant, not a replacement.

I created this Thai Mexican fusion recipe using ChatGPT during the show and might just make it for dinner this week. If I do, I’ll report back on the podcast!

Google Veo 3: From Model to Meme

Google’s Veo 3 has become a meme and movement of its own on platforms like TikTok. Not just the output of the model, but the idea OF the output is catching steam. People are now pretending to be AI-generated avatars, referring to Veo 3, deliberately mimicking the style and quirks of AI-generated clips, and labeling their real videos as “100% AI” to attract attention. I mean, honestly, I’m bummed I didn’t think to do this first! It further blurs the lines of reality and authenticity. Does that damage our ability to discern the differences over time? Or… is it just all about ridiculousness? Oh, and social currency!

Everything you need to know about this show can be found at our site: aiinside.show—subscribe links for your podcatcher of choice, video versions, and live recording details. If you love this show, please leave a review in Apple Podcasts—it really helps!

And if you really REALLY love this show, support us on Patreon for ad-free shows, a Discord community, and an AI Inside T-Shirt by becoming an Executive Producer!

HUGE Thank You to Executive Producers

DrDew, Jeffrey Marraccini, WPVM 103.7 in Asheville NC, Dante St James, Bono De Rick, Jason Neiffer, Jason Brady, and Anthony Downs!!

See you next Wednesday on another episode of AI Inside!