Alibaba swings at US AI lead, Altman wants to pump the brakes
Alibaba drops its most capable model yet, Sam Altman calls for slower AI development, and Congress is quietly running on ChatGPT.
Subscribe free →Alibaba Just Dropped Its Biggest Model and Claimed It Rivals OpenAI
Chinese tech giant Alibaba released its largest open-weight AI model to date — Qwen Max — claiming benchmark performance on par with top US frontier labs including Anthropic and OpenAI. The move is the latest salvo in an increasingly fierce US-China AI race, and the open-weight release means anyone can download and run it. If the benchmarks hold up under scrutiny, this is a serious challenger.
Why it matters: A capable open-weight Chinese model that rivals GPT-4-class systems accelerates global AI proliferation and complicates any US attempts to maintain a technology lead through export controls.
Sam Altman Is Suddenly Talking About Slowing Down AI
The man who built OpenAI into the most influential AI company on earth is now calling on the industry to "pace the rate of AI development." The about-face from Altman — who has spent years racing to build AGI as fast as possible — has sparked a full-blown decel debate inside and outside Silicon Valley. Whether this is genuine caution or strategic positioning remains very much up for debate.
Why it matters: When the CEO of OpenAI starts using deceleration language, it signals either a real safety reckoning or a calculated move to shape incoming regulation — either way, it shifts the conversation.
Congress Runs on ChatGPT — and the Receipts Are Public
House spending records reveal that OpenAI's ChatGPT is the dominant paid AI tool used by congressional offices on Capitol Hill. Staffers are using it to draft memos, summarize legislation, and handle constituent communications. This isn't a pilot program — it's already the status quo.
Why it matters: The fact that the people writing AI laws are themselves deeply dependent on one company's AI product is a conflict of interest that deserves far more attention than it's getting.
The EU's AI Transparency Rules Are Now Live
The EU's AI Act transparency obligations took effect August 2nd, requiring companies to disclose when users are interacting with a chatbot or AI-generated content like deepfakes. The European Commission also released a set of official AI labels companies can use rather than designing their own. Non-compliance now carries real legal risk for any company operating in the EU.
Why it matters: This is the first major AI labeling regime to actually go into force anywhere in the world, setting a template — or a compliance headache — for every global AI company.
AI Cybersecurity Startup Horizon3 Hits $2B Valuation
Horizon3 raised $250 million in a Series E at a $2 billion valuation, betting that the rise of AI-powered attacks has made annual penetration testing obsolete. Companies are now paying for continuous, automated security validation that can keep pace with fast-moving threats. The round reflects a broader wave of enterprise security spending driven directly by AI risk.
Why it matters: AI isn't just creating new products — it's creating new threat surfaces, and the security industry is starting to see valuations that reflect how serious enterprises are taking that.
Benioff-Backed June Wants to Fix Why Most AI Projects Fail
A startup called June came out of stealth with a $20 million pre-seed round backed by Salesforce CEO Marc Benioff, aiming to make AI deployment dramatically simpler for enterprises. The pitch: most companies have bought into AI but can't actually get it into production, and June is building the infrastructure layer to bridge that gap. A $20M pre-seed is a large check for a stealth company, suggesting serious conviction from early backers.
Why it matters: Deployment failure — not model quality — is the number one reason enterprise AI initiatives stall, making the "AI for AI adoption" category a legitimate and growing market.
Quick Hits
- →Wispr Flow's updated terms of service hint it's about to launch a meeting notetaker with summaries and action items, entering the increasingly crowded Otter/Fireflies space — TechCrunch
- →Nuclear startup Valar Atomics raised $1B at a $6B valuation led by Sequoia's Shaun Maguire — the AI datacenter power crunch is making nuclear look very fundable very fast — TechCrunch
- →Home battery startup Base Power raised another $1B to scale production, riding the same AI-driven grid demand wave — TechCrunch
- →Lenovo Googlebook laptops leaked in press images, giving us the first real look at Google's forthcoming laptop OS push — The Verge
AI TLDR