Lot's of interesting stuff this week, including Meta’s Superintelligence shakeup, some big moves (and flops) with AI talent and megabucks, and my review of the Oakley Meta HSTN Limited Edition smartglasses.
But first... massive thanks to this week’s Patron of the Week: Peter! If you’d like to support AI Inside directly and join our community, head to patreon.com/aiinsideshow.
Meta’s Superintelligence Extravaganza
Yann LeCun, Chief AI Scientist and friend of the show, clarified his focus on Meta’s FAIR team, continuing with academic and foundational research, while OpenAI’s Shengjia Zhao now helms Meta’s Superintelligence Lab. That means LaCun works on open, long-term AI science, and Zhao chases fast-moving, riskier tech like Superintelligence. Apple just lost another AI researcher, Bowen Zhang, to Meta, fueling concerns about Apple’s AI momentum if Meta keeps tossing around big money. Meta reportedly tried luring a dozen folks from Mira Murati’s Thinking Machines, including one offer topping $1 billion, but nobody bit! Plus, Zuckerberg’s vision piece on Meta.com is bullish on personal Superintelligence and, yes, glasses for everyone, even if you have perfect vision.
Oakley Meta HSTN Smartglasses: One Week Later
HEY! I got the Oakley Meta HSTN Limited Edition smartglasses! After nearly a week, my take: the style’s loud by design but I honestly don’t mind the look. Fit is tight at the ears, and the Polarized Prizm Ruby lenses are gorgeous. The 4k video recording is heavily processed and vertical-only, so not ideal for my studio purposes unfortunately. AI is a mixed bag as I sort of expected it to be. Sometimes impressive, sometimes confused. Voice is natural and speakers are loud but you might not want to listen to anything while standing inside an elevator with others. Battery life and the case are fantastic. At $500, they’re not cheap. Overall, they are really great for people who are primarily interested in social sharing and basic AI functions. Not so much for my use case, sadly.
Maybe I should get the lower cost Meta Ray-Bans instead. Sales tripled last year, so maybe Zuck’s onto something with his “glasses future.”
Words, Reports, and EMDASHES
Jeff shared a report that looks at how language within LLM output is impacting and influencing the way we talk in the real world. I also hear from people frustrated that words, phrases, or especially EM DASHES are now demonized in AI output.
Trump on AI Copyright and US Leadership
US president Donald Trump attended an AI summit last week, weighing in on the complexity of licensing copyrighted material for model training. He called the idea of payments to every rights holder “not do-able” and warned that strict rules could hinder US leadership in AI. He compared this to China’s more relaxed IP enforcement. If copyright law shifts here, does it signal weaker ownership for creators in general? Something to watch.
Amazon, New York Times, and That News API
Amazon signed a multiyear licensing agreement with the New York Times valued at $20 million per year. This gives Amazon access to news, recipes, and The Athletic, and lets them train models with Times content. $20 million is just 1 percent of the Times’s annual revenue.
The Era of Agentic Browsers
Another week, another browser agent: Microsoft’s Edge is testing Copilot Mode, blending an AI assistant right into your browsing. It can find info, summarize, compare tabs, even manage reservations, although maybe not expertly yet. Reminds me of what Perplexity’s Comet does.
Meanwhile, I checked out the Fellou.ai browser, another agentic browser. OK I didn’t actually USE it, the privacy policies are dirt poor. But its a solid reminder: always read the Terms! Who owns your browsing data, and do you trust them?
Perplexity Mac App Gets MCP
The Perplexity Mac app now integrates MCP (Model Context Protocol) from Anthropic, which connects AI models to all sorts of Mac apps, from Notes, Reminders, Email, to Drive, and more. Setup does seem a bit involved, but this could enable some very interesting Mac workflows. I’m curious to try it out (even though setup is not my favorite thing).
ChatGPT Circumvents CAPTCHAs
OpenAI’s new ChatGPT Agent is so human-like, it bypassed Cloudflare’s “I’m not a robot” test while narrating its actions as it went! Not shocking since CAPTCHAs look for patterns and AI is fantastic at pattern recognition. Personally, I wiggle and pause for CAPTCHAs to look more “human,” but it’s clear these systems need a big rethink.
Google’s AI Mode Expands
Google’s AI Mode adds live video via Lens, PDF and image uploads, a collaborative Canvas, and soon a Chrome address bar chip to “Ask Google about this page.” I’ve been using AI Mode more for fact-checking and I love how fast and comprehensive it has become.
Meta May Allow AI in Technical Interviews
Jason Koebler at 404Media reports Meta is discussing letting job candidates use AI tools during technical interviews. Makes sense: why not let devs use tools everyone’s already embracing? Meta is running internal mock interviews to iron out the details. This shift is happening at schools, too.
Tom Merritt and I talked to Jason Koebler about it on the Daily Tech News Show, check that episode out!
Adobe Photoshop: Generative Upscale and More
Adobe Photoshop has new AI-powered features: Generative Upscale (boosts images up to 8MP, awesome for reviving old photos), Harmonize (matches inserted objects to the scene’s lighting, color, and tone), and smarter object removal. I’m excited to see how effective the upscaling really is. Adobe keeps nailing AI in creative workflows and as someone who lives inside Adobe’s suite, I’m always eager to test the latest.
Runway’s AI Film Festival Hits Imax
Imax is teaming up with Runway to bring the top ten films from Runway’s AI Film Festival to theaters in ten US cities. Out of over 6,000 entries, these short films (2-10 minutes each) represent a wave of new AI art. Part of me cringes at AI films; another part is fascinated and might check them out when they arrive in SF. No denying, we’re at the start of a creative trend that could produce the next generation of visionaries.
HUGE thank you to Executive Producers on Patreon: DrDew, Jeffrey Marraccini, Radio Asheville 103.7, Dante St James, Bono De Rick, Jason Neiffer, Jason Brady, Anthony Downs, and Mark Starcher!!
See you next Wednesday on another episode of AI Inside!