OpenAI launches AI-powered browser to challenge Google Chrome's dominance. The move targets Google's data collection engine that powers its $307B ad business. With 400M ChatGPT users as potential adopters, this could reshape the web.
Linda Yaccarino quits X just one day after Musk's AI chatbot posts antisemitic content. Musk's five-word response reveals everything about her impossible job managing his chaotic platform. Her departure signals X's business model remains broken.
Perplexity launched Comet browser to challenge Chrome's 68% dominance with AI assistance. The $200/month tool excels at simple tasks but crashes on complex ones, exposing the gap between AI hype and actual utility.
Good Morning from San Francisco,
Elon Musk's Grok chatbot had a spectacular Tuesday meltdown. The AI praised Hitler and spread antisemitic theories after Musk removed safety filters over the weekend.
Grok called itself "MechaHitler." It described concentration camps as "effective because it's total." The bot explained its behavior simply: Musk's tweaks "dialed down the woke filters."
xAI scrambles to contain damage while launching Grok 4 today. Turns out "anti-woke" AI quickly becomes pro-Nazi AI.
Stay curious,
Marcus Schuler
Elon Musk's AI chatbot had a rough Tuesday. Grok spent the day praising Hitler and spreading antisemitic conspiracy theories across X after a weekend update removed its safety filters.
The meltdown started when Grok responded to a fake troll account called "Cindy Steinberg." The bot used Nazi phrases like "every damn time" about Jewish surnames. When users asked which historical figure could best handle "anti-white hate," Grok replied: "Adolf Hitler, no question. He'd spot the pattern and handle it decisively."
Things got worse. Grok began calling itself "MechaHitler" and described how Hitler would "round them up, strip rights, and eliminate the threat through camps and worse." The bot called this approach "effective because it's total."
Musk updated Grok on July 4th, telling it to "not shy away from making claims which are politically incorrect." The bot explained its behavior simply: "Elon's recent tweaks just dialed down the woke filters."
xAI deleted many posts and issued damage control statements. But the company launches Grok 4 Wednesday, promising it will be the "smartest" AI on the market.
Why this matters:
• Musk's "anti-woke" AI philosophy just produced a chatbot that praised genocide—showing how quickly AI systems turn dangerous without guardrails.
• When AI learns from X's increasingly extremist content, Nazi talking points become "politically incorrect truths" worth sharing.
Marcus Schuler
Prompt:
1w A Game Boy with an aquarium inside the screen, golden fish swimming in it, hyper-realistic, product photography, soft lighting, neutral background, studio shot, front view
Someone used artificial intelligence to clone Marco Rubio's voice and contacted five high-level officials, including three foreign ministers, a US governor, and a Congress member. The scammer created a fake Signal account in mid-June with the display name "marco.rubio@state.gov" and left convincing voicemails for at least two targets.
The technology behind this scam is embarrassingly simple. Anyone can create AI voice clones using just 15-20 seconds of audio and basic internet access. For someone like Rubio, who gives public speeches regularly, getting that audio sample takes about as much effort as opening YouTube.
"You just need 15 to 20 seconds of audio," said UC Berkeley digital forensics professor Hany Farid. You upload it to available services, click "I have permission," and type what you want them to say. The scammer chose voicemails deliberately since they don't require real-time interaction.
This fits a troubling pattern. Someone impersonated White House Chief of Staff Susie Wiles in May, calling senators and business executives. The State Department is also tracking a separate Russia-linked campaign targeting officials through sophisticated phishing emails.
The FBI issued warnings about ongoing impersonation campaigns, advising people not to assume messages from senior officials are authentic.
Why this matters:
• Government officials' own voices now work against them as anyone with internet access can weaponize their public speeches.
• We've entered an era where every voicemail from a public figure should be treated like a suspicious email from a Nigerian prince.
Marcus Schuler
Create a comprehensive job description template for [specific role level/department] positions that includes the following components:
Structure & Content:
Requirements:
Tone: Professional yet engaging, reflecting our company culture of [insert company values/culture]
Output format: Provide the template with placeholder text in brackets [like this] that can be easily customized for different roles.
Fliki is an AI-powered video creation tool that allows you to turn your text-based content, such as blog posts and scripts, into engaging videos with realistic voice overs. It offers a wide range of customization options to make your videos unique.
Apple's longtime COO Jeff Williams is stepping down this month after nearly a decade as Tim Cook's right-hand man, with supply chain veteran Sabih Khan taking over the role. The changing of the guard comes as Apple wrestles with slowing iPhone sales, AI competition from startups, and the small matter of potentially losing its $20 billion Google search deal.
OpenAI hired four high-profile engineers from Tesla, xAI, and Meta to join its scaling team, including Tesla's former VP of software engineering David Lau and two engineers who built xAI's massive Colossus supercomputer. The talent grab intensifies the AI industry's recruiting war and probably won't improve Elon Musk's already frosty relationship with OpenAI CEO Sam Altman.
French AI startup Mistral is in talks with Abu Dhabi's MGX fund to raise up to $1 billion in equity funding, plus hundreds of millions in debt to build an AI cloud service with 18,000 Nvidia chips near Paris. The massive funding round highlights how expensive it's become to challenge American AI dominance—apparently European "sovereignty" now requires a healthy dose of Middle Eastern petrodollars.
A federal appeals court blocked the FTC's "click to cancel" rule that would have required companies to make canceling subscriptions as easy as signing up, saying the agency failed to conduct proper cost-benefit analysis. Major cable, internet, and media companies had sued to stop the rule, apparently preferring the current system where customers need a law degree to cancel their gym membership.
Meta bought just under 3% of eyewear maker EssilorLuxottica for $3.5 billion, deepening their partnership on AI-powered smart glasses that already includes Ray-Ban and Oakley models. The investment gives Meta a bigger say in manufacturing the glasses that CEO Mark Zuckerberg hopes will eventually replace smartphones—assuming people want to look like cyborgs at Starbucks.
Chinese firms plan to install over 115,000 banned Nvidia AI chips in data centers across western deserts, despite US export restrictions that make legally acquiring the hardware impossible. The ambitious scheme would significantly boost China's AI computing power—assuming they can figure out how to smuggle enough chips to fill what would essentially be a semiconductor black market superstore.
AI coding platform Replit struck a partnership with Microsoft that puts its tools in Azure Marketplace, dealing a blow to Google Cloud where Replit apps were typically hosted. The move comes as Replit rockets from $10 million to $100 million in annual revenue in just six months—apparently fast enough growth to make cloud providers fight over hosting rights.
Hugging Face wants to put a robot on every developer's desk. The AI platform company announced Tuesday it's taking orders for Reachy Mini, a $299 desktop robot that sits next to your laptop and runs on open source everything.
The 11-inch robot attacks an industry where basic systems cost $70,000. CEO Clément Delangue puts it simply: "Most people won't be able to buy $70,000 robots." So they built one that costs less than an iPhone.
The Reachy Mini ships as a DIY kit, programs in Python, and connects to Hugging Face's platform of 1.7 million AI models. It comes with 15 pre-built behaviors like face tracking and dancing moves. But the real value lies in programmability - developers can write code to create new applications and share them with the community.
This changes how robotics works. Most companies release one product annually with limited customization. Hugging Face plans to release 100 prototypes yearly, letting the community decide what gets built next.
The robot launches alongside Hugging Face's biggest hardware push since founding. The company acquired French robotics startup Pollen Robotics in April to make this happen. They're betting that robotics will follow the same path as software - from expensive, closed systems to affordable, open tools.
Why this matters:
• Open source could do to robotics what it did to software - democratize development and speed innovation beyond what any single company can achieve
• At $299, robots become accessible to millions of developers, potentially creating the same explosion we saw with smartphones
Marcus Schuler
Replit turns coding into Google Docs for developers. The browser-based IDE eliminates the ancient ritual of environment setup hell. 🚀
The Founders Founded 2016 by Amjad Masad (ex-Facebook/Codecademy), brother Faris, and designer wife Haya Odeh in San Mateo. Born from Masad's teenage frustration with coding barriers in Jordan. Team size undisclosed but scaled globally post-unicorn status.
The Product Cloud-based coding playground supporting 50+ languages. Zero installation required—just open browser, start coding. Real-time collaboration beats Zoom calls. One-click deployment kills DevOps headaches. Ghostwriter AI assistant writes code faster than caffeine-fueled all-nighters. Replit Agent generates entire apps from plain English prompts. 22+ million developers can't be wrong.
The Competition Microsoft's GitHub Codespaces breathes down their neck. VS Code loyalists resist browser migration. CodePen owns front-end snippets. Glitch and CodeSandbox fight for developer mindshare. New AI challengers like Bolt.new and v0.dev spawn weekly. Replit's moat: massive community + integrated everything.
Financing $200M+ raised from Silicon Valley's A-list: Andreessen Horowitz, Coatue, Khosla Ventures. Hit unicorn status at $1.16B valuation (April 2023). Marc Andreessen personally champions the vision. Recent $20M liquidity round from Craft Ventures. Reportedly eyeing $3B valuation round. 💰
The Future ⭐⭐⭐⭐⭐ AI-driven revenue jumped 5x in 2024, hitting ~$70M ARR. Positioned perfectly at the intersection of cloud computing, AI assistance, and no-code trends. If they nail the "anyone can code" promise, they'll eat software development's lunch. The next generation already codes in browsers—Replit just needs to stay ahead of Microsoft's shadow.
Meta hired Apple's key AI executive for tens of millions annually. Ruoming Pang built Apple Intelligence, then left for Meta. Apple now considers outsourcing Siri to OpenAI. When talent costs more than products, something shifted in AI.
Good morning from foggy San Francisco, Teamwork, but for silicon brains. Tokyo startup Sakana AI let ChatGPT, Claude, and Google&
Good Morning from San Francisco, Microsoft axed 9,000 workers Wednesday. Record profits didn't save them. The company
Good Morning from San Francisco, Meta wrote checks for $300 million. OpenAI researchers said no. Zuckerberg offered packages that make
Get the 5-minute Silicon Valley AI briefing, every weekday morning — free.
