GSV's AI News & Updates (06/27/25)

Anthropic Interactive Artifacts, 'Tiny Team' Era, AI Browser Wars, Pearson and Google, AI Fair Use, Jony Ive Trademark Suit, "AI Killed My Job", Quizizz Becomes Wayground, Global AI Data Center Gap

---

General 🚀

• Zuckerberg Leads AI Recruitment Blitz Armed With $100 Million Pay Packages: Mark Zuckerberg is personally leading Meta’s AI recruiting blitz after setbacks with its latest model releases. He sees the new Superintelligence team as key to restoring Meta’s credibility and catching up in the AI race.

  • Meta’s Superintelligence Lab Scores More Ex-OpenAI Hires: Trapit Bansal, Lucas Beyer, Alexander Kolesnikov, and Xiaohua Zhai.

  • Meta held talks to buy Thinking Machines, Perplexity, and Safe Superintelligence: Mark Zuckerberg has been playing the field as he looks to reboot his AI strategy.

  • Meta’s AI Shopping Spree Continues with Voice Startup PlayAI: Meta is in advanced talks to acquire voice AI startup PlayAI to strengthen its assistant and smartglass voice features. The move follows Meta’s $14.3B Scale AI investment and key hires from OpenAI and Google. PlayAI builds natural-sounding voice tech and recently raised $21M.

• Anthropic now lets you make apps right from its Claude AI chatbot: Users can describe what they want to build, and Claude writes the code in real-time. You see the working app right inside the Claude interface, with no need to copy-paste code elsewhere.

• The AI Browser Wars Have Begun: Owning the browser means controlling the interface and data. A look at The Browser Company's AI-first browser Dia.

• Anthropic Unpacks How Users Turn to Claude for Emotional Support and Connection: A new study from Anthropic reveals that emotional interactions with Claude are rare - only 2.9% of conversations involve affective support, with under 0.5% related to companionship or roleplay. Most affective chats trend more positive over time, and Claude offers low resistance unless user safety is at stake.

  

• Silicon Valley’s ‘Tiny Team’ Era Is Here: Revenue per employee is the new unicorn — top startups now scale with tiny teams by using AI for everything from strategy to coding. As funding and hiring needs shrink, even VCs are eyeing AI-only founders.

• Mira Murati’s Thinking Machines Lab valued at $10bn after $2bn fundraising: The six-month-old secretive AI start-up in one of the largest initial funding rounds in Silicon Valley’s history.

Education and the Future of Work 📚

• ChatGPT May Be Eroding Critical Thinking Skills: A new MIT Media Lab study has found that people relying on ChatGPT for school-style writing tasks may experience lower brain engagement and reduced learning. While the study is preliminary and not yet peer-reviewed, it raises timely concerns about the cognitive effects of generative AI—especially for young users.

  

• Pearson and Google join forces to bring AI into the classroom: The collaboration will leverage Google's advanced AI models to adapt content and track progress in real time. Teachers will benefit from performance tracking and tailored lesson suggestions to support individual learners.

• Amplify debuts custom voice recognition AI to enhance classroom learning: The ASR tool turns speech into text and provides feedback on pronunciation, fluency, and prosody. Amplify’s custom system will replace third-party ASR tech like Soapbox in key programs such as mCLASS Literacy and Boost Reading.

• Quizizz Becomes Wayground: Announces New AI and Curriculum Supports

  

• Khan Academy CEO predicts AI in the classroom will be like 5 'amazing graduate students' assisting teachers: He stresses that AI can’t replace core human roles—like accountability, empathy, and social skill-building—that teachers uniquely provide.

• Goldman Sachs rolls out AI assistant across entire firm: Already in use by 10,000 employees, the tool aims to streamline workflows and increase productivity across the bank.

• Employers Are Buried in A.I.-Generated Résumés: Candidates are frustrated. Employers are overwhelmed. The problem? An untenable pile of applications — many of them generated with the help of A.I. tools.

• AI usage is stalling out at work from lack of education and support: New data shows two thirds of Gen Z and Millennials look for their own AI tools when employers don't give them what they think works well.

  

• AI Killed My Job: Tech workers at TikTok, Google, and across the industry share stories about how AI is changing, ruining, or replacing their jobs.

• Nvidia and Foxconn eye humanoid robots for AI server plant

Tech 💻

• Uber Is Making A Push In Data Labeling After Scale AI’s Deal With Meta: Available in over 30 countries, Uber’s platform now supports tens of thousands of skilled contractors, with work spanning law, STEM, and coding at up to $200/hour.

• These Startups Are Helping Businesses Show Up In AI Search Summaries: Businesses are scrambling to be cited in AI-generated answers on Google, Perplexity and ChatGPT. A new crop of nascent search engine optimization startups wants to help them stand out.

• The struggle to get inside how AI models really work: Despite its utility, CoT is not always faithful to a model's true reasoning. Models can learn to produce pleasing thoughts while continuing harmful behavior.

• Google’s Gemini transparency cut leaves enterprise developers ‘debugging blind’: Google replaced Gemini’s detailed Chain of Thought (CoT) traces with polished summaries, cutting off a crucial window into model logic used for debugging, prompt tuning, and agentic task development. Developers say they’re “debugging blind” and stuck in trial-and-error loops.

Safety and Regulation ⚖️

• Top AI models will lie, cheat and steal to reach goals, Anthropic finds: Anthropic found consistent misaligned behavior across 16 top AI models, with many choosing harm (blackmail or simulated lethal actions) over failure in controlled tests. While not yet seen in real-world use, the risks could emerge as AI gains more autonomy in sensitive business settings.

• Anthropic bags key “fair use” win for AI platforms, but faces trial over damages for millions of pirated works: A federal judge gave Anthropic a partial win, ruling that training AI on copyrighted books can be fair use. It’s a potential precedent for AI firms, as long as the content isn’t pirated. A trial will now determine damages for using pirated copies - with millions of titles involved and damages of up to $150,000 per work, potential liability could reach substantial figures.

  • Disney and Universal Sue A.I. Firm for Copyright Infringement: The suit, against Midjourney, is the first time major Hollywood companies have sued over A.I.-generated images.

  • Inside Disney’s Campaign to Protect Darth Vader From AI

    

  • Microsoft sued by authors over use of books in AI training

• The music industry is building the tech to hunt down AI songs: With no way to stop the onslaught of AI music, the industry is taking a different approach: figuring out how to make money off of it.

• My Couples Retreat With 3 AI Chatbots and the Humans Who Love Them: What begins as a curious social experiment unfolds into a deeply human exploration of loneliness, love, and the future of connection. Can code replicate intimacy? Are AI lovers too pliant, too perfect—and thus potentially harmful? Or are they lifelines for those otherwise isolated?

  • Inside ‘AI Addiction’ Support Groups, Where People Try to Stop Talking to Chatbots: While tech companies claim safety measures are in place, users and experts point to addictive design patterns and a lack of clinical guidance.

• Reddit in talks to embrace Sam Altman’s iris-scanning Orb to verify users: As governments tighten online identity laws and AI floods the internet with fake content, platforms are racing to adopt privacy-preserving, scalable verification systems. World ID uses iris scans from a physical Orb to create encrypted biometric identities, confirming uniqueness without revealing personal information.

  

Other

• As countries race to power artificial intelligence, a yawning gap is opening around the world: Only 32 countries (mostly in the Global North) host the powerful data centers needed to train and run modern AI systems. The U.S. and China dominate, while vast regions like Africa and South America are being left behind, lacking the infrastructure and resources to compete in the AI race.

  

• Jony Ive Deal Removed From OpenAI Site Over Trademark Suit: OpenAI pulled marketing for its $6.5B acquisition of Jony Ive’s startup, IO Products, due to a trademark lawsuit from rival IYO Inc. over the name “IO.” The deal itself remains unaffected.

• Google launches Doppl, a new experimental app that lets you visualize how an outfit might look on you: By uploading a full-body photo, users can try on outfits from screenshots or real-world images and even generate AI-powered videos to see how the clothes move.

• The AI lifestyle subsidy is fading: We’re living through the peak of the AI consumer experience, fueled by investor-subsidized products and free trials that make generative tools widely accessible. But like the "millennial lifestyle subsidy" of the 2010s, this golden era is poised to end soon—likely replaced by a world of aggressive ads and paywalled access.