Good Morning from San Francisco,

Tech giants want AI regulation - but only on their terms.

👉 Google demands free use of copyrighted content for AI training, while OpenAI pushes to ban Chinese competitors. Their reason? DeepSeek's founder once met Xi Jinping. That's like calling someone a coffee addict for visiting Starbucks once. ☕

👉 Meanwhile, Y Combinator breaks ranks. The startup kingmaker backs Europe's tech crackdown, pushing the White House to adopt EU-style rules. They've invested $2 million in political action. Small change, big statement. 🔄

The irony bites. Silicon Valley's "move fast and break things" crowd now begs for regulation - as long as it hurts their rivals. 🎭

Stay curious,

Marcus Schuler

Silicon Valley's Two-Faced AI Strategy Revealed

Google and OpenAI just showed their cards. They want AI rules that help them crush competition while dodging oversight.

Google thinks all public data should be free lunch for AI training. Their new proposal would let them use copyrighted content without paying creators. Convenient timing - it matches the Trump administration's work on AI strategy.

Meanwhile, OpenAI launched a surprise attack on Chinese AI lab DeepSeek. They're pushing for a U.S. ban on "PRC-produced" AI models. Their evidence? DeepSeek's founder once met Xi Jinping. That's like calling someone a coffee addict because they visited Starbucks.

The irony is thick. Google fights regulation at home but begs for government help abroad. OpenAI wants to block Chinese competition while expanding its market power. Both companies act shocked when anyone suggests limiting their own behavior.

States aren't buying it. With 781 AI bills pending, lawmakers are ready to act. Google's response? "Trust us, we're innovating." A classic move from the Silicon Valley playbook.

Why this matters:

• Tech giants want to write AI's rules - imagine letting foxes design henhouse security
• Their "proposals" are thinly disguised power grabs that would reshape who controls AI's future

Read on, my dear:

• OpenAI#s Policy Proposal
• Google's Policy Proposal

Silicon Valley Rebels: YC Pushes White House to Embrace EU Tech Rules

Y Combinator just picked a fight with Silicon Valley's giants. The startup accelerator wants the White House to support Europe's Digital Markets Act – a move that's raising eyebrows across the tech world.

European regulators crafted the DMA to tame six tech behemoths: Apple, Google, Meta, Amazon, Microsoft, and ByteDance. The law forces them to play fair with smaller competitors. YC thinks that's exactly what American startups need.

The timing packs extra punch. Apple just pushed its AI-powered Siri upgrade to 2027. YC pounced on this delay as proof that tech giants need competition to innovate. They're tired of watching promising startups get crushed under Big Tech's thumb.

YC isn't just talking. They've put $2 million into political action lately. That's pocket change compared to Andreessen Horowitz's $89 million, but it signals serious intent. The accelerator wants Washington to know they mean business.

Why this matters:

• Silicon Valley's "move fast and break things" crowd now wants regulation – as long as it breaks up their bigger rivals
• The startup world's newfound love for European-style tech rules marks a seismic shift in the industry's power dynamics

AI Photo of the Day

Prompt:
cute, ethiopian baby, crawling, side view, no background, white floor tile

Maya Maker Goes Open Source: AI Voice Clone Tech Now Free

Sesame, the startup behind viral AI assistant Maya, just released its voice generation model to the public. The billion-parameter CSM-1B model can clone voices in under a minute with virtually no safeguards.

The tech uses Meta's Llama as its foundation and employs "residual vector quantization" to turn text into eerily human-like speech. It's now available under an Apache 2.0 license, meaning anyone can use it commercially.

Testing revealed the model's concerning capabilities. I cloned my own voice rapidly and generated speech on sensitive topics like elections and propaganda. The only protection? An honor system asking users to behave.

Meanwhile, Sesame isn't just about voices. The Oculus co-founder's startup is developing AI glasses for all-day wear, backed by major VC firms.

Why this matters:

• We've entered an era where anyone can clone voices at will - consumer protection hasn't caught up
• The honor system approach to AI safety looks increasingly naive as powerful models go public

Read on, my dear:

• Sesame Website

AI & Tech News

iPhone Users Get Secure Android Messaging at Last

Apple just ended the blue vs. green bubble war by adding encrypted RCS messaging support across its devices. In a move that has messaging nerds everywhere dropping their phones in disbelief, the tech giant will finally let iPhone users securely text their Android friends without feeling like second-class citizens.

Foxconn Boss: Trump's Tariffs Give Tech Giants Migraines

Foxconn's CEO Young Liu just aired Silicon Valley's dirty laundry: Trump's tariff threats are driving Apple and Amazon executives to pace their corner offices. The world's biggest electronics manufacturer, caught between Chinese factories and American politics, watches its customers scramble to cook up U.S. manufacturing plans faster than you can say "trade war."

China Joins AI Content Crackdown

China's internet watchdog ordered all AI-generated content to carry warning labels starting September 1. The move aligns Beijing with Western regulators, who apparently agree on something for once.

Tinder Parent Gets Reality Check on Dating Game

Match Group's new CEO Spencer Rascoff admits the company's apps have become a shallow numbers game instead of fostering genuine relationships. In a candid LinkedIn post that reads like a dating profile do-over, Rascoff promises to "streamline operations" and create confidential channels for employees to help fix the love-finding formula.

SoftBank Turns TV Factory Into AI Powerhouse

SoftBank just dropped $676 million on Sharp's old LCD panel factory in Osaka, planning to transform it into a massive AI data center that will power its partnership with OpenAI. The tech giant's shopping spree highlights an oddly poetic transition in Japan's industrial landscape: where workers once crafted screens for humans to watch, machines will now crunch data to think like them.

BMW's Profit Drop: Not As Bad As It Looks

BMW's profits tumbled 37 percent to 7.7 billion euros, but industry experts say this reflects a return to normal after several extraordinary years. Despite challenges in China and brake supply issues, the automaker maintains an enviably profitable position that most companies would kill for.

AI decoded 🔓: AI self-hosted

AI Power Without the Price Tag: Build Your Own AI Hub

A new open-source project lets you access ChatGPT, Claude, and other AI models from a single interface - without paying for multiple subscriptions.

OpenWebUI, paired with LiteLLM, creates a personal AI hub that runs on anything from a laptop to a cheap cloud server.

The setup process is surprisingly simple. Install OpenWebUI on a virtual private server, add your API keys, and you're running multiple AI models side by side. It comes pre-loaded with Llama, while cloud giants like GPT-4 and Claude plug in seamlessly.

The cost savings are significant. Instead of monthly subscriptions, you pay per token used. Light users spend under $5 monthly accessing premium AI models. Power users can mix and match models to optimize their budget.

LiteLLM handles the backend magic, automatically routing queries to the most cost-effective AI model available. If one service goes down, it smoothly switches to alternatives. Think of it as your personal AI traffic controller.

The system works equally well for individuals, families, or businesses. Parents can monitor kids' AI usage. Companies can track employee access. Everyone benefits from improved privacy since queries route through your own server.

Why this matters:

• You're not locked into expensive subscriptions. Use any AI model, pay only for what you need
• Finally, a way to wrangle multiple AI services without losing your mind (or emptying your wallet)
  

🚀 Quick Setup Guide: Your Personal AI Command Center

1. Get Started

• 🖥️ Rent a basic VPS from Hostinger ($5-10/month)
• 📦 Choose Ubuntu as your operating system
• 🔑 Save your server login details

2. Install Core Software

• 👾 Update Ubuntu: apt update && apt upgrade
• 🐍 Install Python: apt install python3-pip
• 🔧 Get Docker: curl -fsSL https://get.docker.com | sh
  

3. Set Up OpenWebUI

• 📥 Pull the image: docker pull openwebui/open-webui
• 🏃‍♂️ Run it: docker run -p 8080:8080 openwebui/open-webui
• 🌐 Access at: http://your-server-ip:8080
  

4. Add Your AI Keys

• 🔐 Get API keys from OpenAI, Anthropic, or Google
• ⚡ Paste them in OpenWebUI's settings
• 💳 Remember: You only pay for what you use
  

5. Connect LiteLLM

• 🤖 Install: pip install litellm
• 🔄 Set as default router in OpenWebUI
• 🎯 Let it auto-select the best AI model
  

Done! You now have your own AI hub. Total setup time: About 30 minutes.
Tip: Start with GPT-3.5 - it's cheap and fast. Add fancier models later.