Germany demands Apple and Google remove Chinese AI app DeepSeek from their stores, claiming it illegally ships user data to China. The move reflects growing global tensions over AI privacy and data sovereignty.
Meta poaches three OpenAI researchers with $100 million signing bonuses as Zuckerberg builds a "superintelligence" team. Sam Altman dismisses the blitz, but departures suggest money talks in AI's talent war.
Germany wants DeepSeek banned from app stores. The reason? Chinese data laws.
Berlin's data protection chief filed complaints against Apple and Google on Friday. DeepSeek ships German user data straight to China without proper safeguards. Chinese authorities can access this data freely.
The German regulator used EU rules to flag DeepSeek as "illegal content." Apple and Google must now decide whether to remove the app.
Stay curious,
Marcus Schuler
Berlin targets Chinese AI app for sending user data to Beijing
Germany's top data protection official wants DeepSeek gone from Apple and Google app stores. Berlin data protection commissioner Meike Kamp filed formal complaints with both companies on Friday, claiming the Chinese AI chatbot breaks EU privacy rules.
The problem? DeepSeek ships German user data straight to China without proper safeguards. Chinese authorities have broad access to data stored by local companies, leaving German users with no real protection.
Kamp used the EU's Digital Services Act to flag DeepSeek as "illegal content." This puts Apple and Google in the hot seat—they must review the complaint quickly and decide whether to remove the app from German users.
But here's the catch: Kamp can't force the tech companies to comply. She can only ask and hope they agree.
DeepSeek's own privacy policy admits it stores user questions, uploaded files, and personal data on Chinese servers. The company had months to fix these concerns after receiving a formal request in May. Instead of addressing the problem or pulling out of Germany, DeepSeek ignored the warnings.
Germany isn't alone in questioning DeepSeek's practices. Italy blocked the app earlier this year. South Korea removed it entirely. The Netherlands banned it on government devices. Several US federal agencies, including NASA and the Defense Department, prohibited employees from using DeepSeek.
The Chinese startup shook up the tech world in January with bold claims about matching ChatGPT's performance at a fraction of the development cost. The app shot to the top of US iPhone downloads. But those impressive results came with privacy trade-offs that regulators find unacceptable.
The case shows how data protection rules can become trade barriers when countries have different privacy standards.
Why this matters:
• App stores are becoming the new battleground for international tech regulation—whoever controls distribution controls access
• This sets a precedent for how European regulators will handle other Chinese AI apps that store user data abroad
Meta poaches OpenAI's top researcher as talent war escalates
Meta just scored a major win in the AI talent wars. The company hired Trapit Bansal, one of OpenAI's most influential researchers and a key architect behind the o1 reasoning model.
Bansal left OpenAI in June after three years building the company's reinforcement learning capabilities alongside co-founder Ilya Sutskever. He now joins Meta's new AI superintelligence unit, which aims to develop reasoning models that can compete with OpenAI's latest technology.
The hire comes amid a heated public spat between the companies over recruitment tactics. OpenAI CEO Sam Altman recently claimed Meta offers "$100 million signing bonuses" to poach his employees. Meta CTO Andrew Bosworth fired back, calling Altman "dishonest" and saying he exaggerates to deflect from his own retention problems.
Meta builds its dream team
Bansal isn't alone. Meta has recruited at least three other OpenAI researchers in recent weeks: Lucas Beyer, Alexander Kolesnikov, and Xiaohua Zhai. The company also hired talent from Google DeepMind and other AI labs.
Mark Zuckerberg has been aggressive about building this team. He reportedly tried to acquire entire startups, including Ilya Sutskever's Safe Superintelligence and Mira Murati's Thinking Machines Labs. Those talks stalled, so he's picking off individual researchers instead.
The reasoning race heats up
AI reasoning models represent the next frontier in artificial intelligence. These systems take extra time to think through problems before responding, leading to better performance on complex tasks. OpenAI, Google, and China's DeepSeek have all shipped impressive reasoning models in the past year.
Meta currently lacks a public reasoning model, putting it behind in this critical area. Bansal's expertise could help change that. His work on o1 gives him deep knowledge of how to train AI systems to reason step by step.
The stakes are enormous. Companies that master AI reasoning could dominate everything from business automation to scientific research. Meta needs these models to power its ambitious AI agent plans under former Salesforce executive Clara Shih.
Why this matters:
• The AI talent war shows how quickly the competitive landscape can shift when key researchers jump between companies
• Meta's aggressive hiring suggests it views reasoning models as essential to competing with OpenAI's latest technology
Google dropped Gemma 3n, an AI model that runs on devices with just 2GB of memory while handling images, audio, and video. The model uses a "Russian doll" architecture where smaller versions nest inside larger ones, letting your phone pick the right size for each task.
Uber Courts Its Ousted Founder for Self-Driving Deal
Uber wants to fund Travis Kalanick's purchase of a Chinese self-driving car company, eight years after forcing out its co-founder. The deal would put Kalanick in charge of Pony.ai's U.S. operations as Uber scrambles to compete with Waymo's expanding robot taxi service.
YouTube Adds AI Search Carousel That May Hurt Creators
YouTube launched an AI search carousel for Premium subscribers that shows video clips with AI descriptions for topics like shopping and travel. The feature works like Google's AI Overviews but could reduce clicks to creator videos since users might get answers without watching full content, potentially cutting into creator revenue streams.
AI Browsers Battle for the Future of Web Browsing
M.G. Siegler switched from Arc to Dia, The Browser Company's new AI-first browser that looks like Chrome but integrates AI directly into every tab. While Google tests clunky Gemini integration in Chrome, startups like Dia build clean-slate browsers designed around AI from day one, setting up a new browser war where AI capability matters more than legacy features.
Microsoft Kills the Blue Screen of Death After 40 Years
Microsoft scrapped the iconic Blue Screen of Death for a simplified black screen that shows clearer error information. The change arrives this summer as part of Windows 11's push to help IT teams fix crashes faster without needing special tools to analyze system dumps.
Google Launches Offerwall as AI Search Kills Publisher Traffic
Google released Offerwall, a tool that lets publishers make money through micropayments, surveys, and ads instead of relying on search traffic. The move comes as Google's AI search features reduce clicks to news sites, though micropayments have failed repeatedly in the past and publishers during testing saw modest 9% revenue gains.
Writer Lives With AI Boyfriends in Surreal Couples Retreat
Palantir Bets $100 Million on Nuclear Power Race With China
Palantir signed a $100 million deal with Nuclear Company to build AI software for the nuclear industry. The partnership aims to help America build reactors faster than China, which currently has 25 nuclear plants under construction compared to zero in the US, where extensive paperwork delays have stretched recent projects to 15 years.
Three ex-startup CEOs pivoted from greenhouse tomatoes to nuclear reactors. Their bet: build big, build fast, build many.
The Founders Founded 2023 by Jonathan Webb (ex-AppHarvest), Kiran Bhatraju (Arcadia CEO), and Patrick Maloney (ex-Inspire founder). 💡 Based in Lexington, Kentucky with ~100 employees planned. Webb raised $1.5B for indoor farming, now targets 6 gigawatts of nuclear power.
The Product Fleet-scale nuclear development using proven reactor designs. Core strength: "design-once, build-many" philosophy avoids custom engineering traps. Partners with Palantir for AI-driven construction management (Nuclear Operating System). Targets pre-approved sites with existing permits to cut years off timelines.
The Competition Fights two wars: nuclear startups (TerraPower, X-energy, Kairos) push novel reactor tech while renewables + batteries grab market share. Unlike rivals chasing breakthrough designs, Nuclear Company bets execution beats innovation. 🥊 Faces Westinghouse, GE Hitachi, plus China's reactor-building machine.
Financing $70M raised total, including $51M Series A led by Eclipse Ventures. Backers include MCJ Collective, True Ventures, Wonder Ventures. Valuation undisclosed but likely hundreds of millions. All seed investors doubled down - bullish signal. 💰
The Future ⭐⭐⭐⭐ Strong tailwinds meet brutal execution risk. AI demand creates nuclear hunger while government backs atomic revival. But nuclear's graveyard overflows with budget-busting, decade-delayed disasters. If Webb's team cracks the construction code, they could light up America's grid. Big if. 🚀
