The Comunicano for Friday May 2nd 2025

Fast Friday

Remember when the App Store was Apple’s private kingdom? A walled garden with a velvet rope and a 30% cover charge? Well, it seems the velvet's worn thin. After a bit of legal arm-twisting and a whole lot of developer grumbling, Apple just gave Spotify—and everyone else—the keys to the side door. You can now pay for music, podcasts, and probably your existential dread outside the Apple ecosystem. No more giving Tim Cook a third of your sandwich just for biting into it.

Meanwhile, MoviePass—yes, that MoviePass—is back from the dead. This time, it wants you to be a fantasy film exec. Predict box office hits and win imaginary money. It’s like fantasy football for people who quote Wes Anderson movies at brunch.

And out in Texas, trucks are driving themselves. No hands, no feet, just steel and code barreling down I-45 like it's just another Tuesday. Somewhere, a long-haul driver just took up podcasting full-time.

As for the AI world? Let’s just say the future is arriving fast—and expensive. OpenAI retired GPT-4 like an aging rock star, only to drop a $20K/month version with a PhD and a caffeine addiction. Microsoft is printing money with cloud AI, Anthropic wants Claude to be your new best friend, and Pinterest is now the unexpected moral compass of the internet, bravely swatting away AI-generated glitter vomit.

And in the most surreal chapter: Nvidia and Anthropic are feuding over whether Chinese smugglers are hiding chips in prosthetic bellies and boxes of lobsters. Yes, lobsters. In 2025, that's a real sentence.

The moral? In tech, as in life, the lines between genius, nonsense, and the future are thinner than ever. And to dance that tightrope, you read THE COMUNICANO!!!

Andy Abramson

Apple Watch

Apple Opens App Store to External Payment Links (9to5Mac)—Apple has revised its App Store guidelines, now allowing U.S. developers to include external payment links inside their apps. This change follows a legal ruling in the Epic Games case that found Apple in violation of a 2021 injunction. Apps can now feature buttons or links leading users to outside purchasing platforms without special entitlements. Importantly, Apple cannot charge commission fees or restrict how these links are presented. The move marks a pivotal shift toward more competition and developer flexibility within the iOS ecosystem. Read more here

Spotify Unlocks External Payments on iOS After Apple Rule Shift (The Verge)—Spotify has rolled out an iOS app update that finally lets U.S. users pay for subscriptions outside Apple's payment system. This follows Apple’s revised App Store rules, allowing external payment links without platform fees. Spotify’s update enables users to view pricing, access special deals, and switch plans directly. This change is a big win for app developers looking to bypass Apple’s longstanding 30% cut, and it signals a broader shift in how mobile apps can handle commerce. Read more here

Entertainment Watch

MoviePass Launches Fantasy Platform for Film Buffs Called ‘Mogul’ (TechCrunch)—MoviePass is reinventing itself with 'Mogul,' a daily fantasy game for movie fans. Still in beta, it lets users act as virtual studio executives, predicting box office numbers, critic scores, and award outcomes using fantasy-style mechanics. The platform integrates blockchain to manage digital rewards and in-game currency. Over 400,000 users have already signed up for early access. With its unique spin on film fandom, MoviePass aims to blend entertainment and gamification into a new kind of audience engagement. Read more here

AV Watch

Aurora’s Driverless Trucks Begin Freight Deliveries in Texas (The Verge)—Aurora has started operating fully autonomous Class 8 trucks on highways in Texas, making deliveries between Dallas and Houston with no human driver on board. The trucks have logged over 1,200 miles for commercial clients including Uber Freight and Hirschbach. Aurora plans to expand this driverless network to include El Paso and Phoenix by year’s end. The milestone reflects growing confidence in autonomous logistics, with a focus on safety, reliability, and real-world deployment of AI-powered transport. Read more here

OpenAI Watch

GPT-4: Gone Too Soon, But Not Forgotten (The Neuron Daily)—OpenAI has officially retired GPT-4 after just two years – a model that transformed entire industries and sparked an AI arms race. While the legacy continues through GPT-4o and anticipated GPT-5, OpenAI faces staggering financial challenges. They're burning $2.25 for every $1 earned, projecting to lose $14B+ in 2025 alone. Meanwhile, they're paradoxically launching a $20K/month doctorate-level AI system using their o3 and o4 models, promising Tesla-like scientific breakthroughs for corporations willing to pay this premium price. Economic reality may burst this bubble before customers can benefit. Read more here

AI Watch

Claude Gets Social: AI Assistant Now Playing Well With Other (Anthropic)—Anthropic has launched "Integrations," empowering Claude to connect directly with your favorite apps and tools. This new capability allows Claude to access and manipulate data across platforms including Jira, Confluence, Zapier, and PayPal. The system builds on their Model Context Protocol (MCP) standard, expanding Claude's usefulness beyond conversation into action. Additionally, they've enhanced their Research capabilities with an advanced mode that searches across the web, Google Workspace, and connected apps, delivering comprehensive reports with proper citations. Both features are available in beta for Max, Team, and Enterprise subscribers, with Pro access coming soon. Read more here

Pinterest's War on 'AI Slop': Fighting Back Against Generated Content (Techcrunch)—Pinterest has launched new tools to identify and filter AI-generated content following mounting user complaints about "AI slop" overwhelming the platform. The visual discovery network, once known for its curated human-created content, has seen an influx of low-quality, algorithmically-produced images that threaten the user experience. Their new detection systems aim to prioritize authentic, high-quality content while filtering out mass-produced AI creations lacking originality or value. The move responds to growing concerns across social platforms about maintaining content quality standards as generative AI tools become increasingly accessible to creators and spammers alike. Read more here

Microsoft Clouds the Competition with AI Revenue Surge (Quartz)— Microsoft's latest earnings report demonstrates that AI investments are paying substantial dividends as cloud revenue from Azure grows by 33%, with nearly half attributed directly to AI workloads. This performance validates the massive capital expenditures on AI infrastructure that many had questioned. Enterprise demand is accelerating rapidly with commercial bookings jumping 17% and future revenue backlog exceeding $315 billion. Major clients including Coca-Cola, BNY Mellon, and Abercrombie are embracing Microsoft's AI solutions, while the company reports its infrastructure is becoming more efficient, converting massive capital investments into returns faster than anticipated. Read more here

AI Race Tightens: US Lead Shrinking as China Catches Up (Spectrum.IEEE)—Stanford University's comprehensive 2025 AI Index reveals the global AI competition is heating up dramatically. While the US still leads in quantity of notable AI models, Chinese systems are rapidly closing the quality gap. The performance difference on benchmark tests has shrunk from 9.26% in early 2024 to a mere 1.70% by February 2025. Many traditional benchmarks have become "saturated" as AI systems achieve near-perfect scores, rendering them increasingly useless for measuring progress. Meanwhile, data access challenges mount as 48% of top web domains now fully restrict bots from crawling their sites, potentially limiting future training capabilities. Read more here

Salesforce Diagnoses 'Jagged Intelligence': The AI Consistency Problem (Venture Beat)—Salesforce has unveiled new tools and benchmarks aimed at addressing the "jagged intelligence" problem plaguing AI systems in business environments. This phenomenon occurs when AI demonstrates unpredictable performance despite high capabilities on paper, causing frustration for enterprise users. Their solution provides standardized measurements and improvements for consistency across business-specific tasks. The initiative comes as companies increasingly rely on AI for critical business functions but face challenges with reliability. Salesforce's approach focuses on creating more dependable AI experiences through rigorous testing and targeted enhancements to reduce performance variability. Read more here

Do AI Models Really Understand Language? (Quanta Magazine)—Computer scientist Ellie Pavlick tackles one of AI’s big philosophical questions: does AI understand language or just simulate it? She argues that even the people who build large language models don’t fully know how their systems work. Despite complete access to the model architecture and training data, researchers often can’t explain why models respond in specific ways. This ambiguity raises questions about interpretability, trust, and the essence of intelligence itself—especially as these systems become more embedded in daily life. Read more here

AI Reshapes Work: MIT Finds 50% Job Reduction in Case Study (MIT Sloan)—A revealing MIT study highlights how a financial services firm used AI to restructure work—leading to major changes. The company broke down jobs into tasks, redeployed AI where possible, and rebuilt roles around human strengths. The results were stark: 50% reduction in workforce, 18% lower employee turnover, and 40% cost savings. Yet, customer satisfaction and outcomes improved, showing how strategic AI deployment can cut costs while enhancing service. It’s a case study in digital transformation with real-world consequences for jobs and efficiency. Read more here

Google Watch

Gemini Evolves: Google Adds Image Editing to Its AI Chatbot (The Verge)—Google’s Gemini chatbot just got more powerful, now offering image editing tools for both AI-generated and user-uploaded visuals. Users can apply multi-step changes, fine-tune prompts, and benefit from the added security of SynthID watermarks. This move positions Gemini to compete directly with Adobe and Canva in the creative space, with an AI twist that simplifies complex editing tasks for everyday users. The layered editability gives more creative control than typical AI image generators, expanding its usefulness across content creation, marketing, and design. Read more here

Microsoft Watch

Microsoft Pushes Passwordless Future with Default Passkeys (The Verge)—Microsoft is phasing out passwords by making passkeys the default sign-in method for new accounts. Users are encouraged to use secure alternatives like biometric logins, push notifications, and physical security keys. Backed by a redesigned sign-in interface, this shift aims to enhance security and usability. Microsoft reports a 98% login success rate with passkeys, compared to only 32% with traditional passwords. Nearly a million passkeys are created daily, signaling rapid adoption. Reflecting the change, Microsoft has rebranded “World Password Day” to “World Passkey Day,” committing to broader support in the year ahead. Read more here

Chip Watch

Chip Wars: Nvidia Accuses Anthropic of Lobster-Gate Fabrications (CNBC)—A rare public clash has erupted between tech giants Nvidia and Anthropic over US chip export policies. Anthropic advocated for stricter controls on AI chip exports to China, claiming smugglers use "prosthetic baby bumps" and "live lobsters" to transport restricted hardware. Nvidia fired back, calling these claims "tall tales" and asserting that "America cannot manipulate regulators to capture victory in AI." The controversy centers around the "AI Diffusion Rule" set to take effect May 15, which would limit advanced chip exports. The dispute highlights growing tensions between hardware providers and AI model developers as competition with China intensifies. Read more here

Robot Watch

Here Comes The Hive (Live Science)—A team from UC Santa Barbara and TU Dresden has created a swarm of 30 compact robots that can flow like a liquid and then harden to support the weight of a person. Each robot, 2.75 inches in diameter, uses magnets and gears to attach and move with others in the swarm. This allows them to shift between fluid and solid states, bearing loads up to 700 newtons—500 times their own weight. Inspired by how embryonic tissue behaves, the robots can self-organize into honeycomb-like structures and change form dynamically, opening up possibilities in adaptive materials and robotics. Read more here