OpenAI and Anthropic both file for IPOs as AI power race heats up
Both AI giants race to go public, Anthropic drops its most powerful model yet, and Apple finally shows up to the AI party at WWDC.
Subscribe free →Both OpenAI and Anthropic Are Racing Each Other to the Public Markets
OpenAI filed confidentially for an IPO, following Anthropic's own filing just days earlier — and the timing is almost certainly not a coincidence. With both companies burning through cash at historic rates and eyeing massive valuations, the public markets are about to get their first real stress test of whether frontier AI is worth the hype.
Why it matters: Two companies that collectively define the frontier AI landscape going public in the same window will set the pricing benchmark for the entire AI industry.
Anthropic Releases Claude Fable 5 — Its Most Powerful Model Available to the Public
Anthropic just made its first Mythos-class model publicly accessible with Claude Fable 5, calling it exceptional at software engineering, knowledge work, and vision tasks. The model ships with hard guardrails blocking responses in high-risk domains like cybersecurity and biology — a deliberate nod to Anthropic's own recent warnings that AI is getting genuinely dangerous.
Why it matters: Releasing a Mythos-class model publicly, even a guardrailed one, marks a meaningful escalation in what's available to everyday developers and researchers.
Apple's Siri Overhaul at WWDC: Catching Up Has Never Looked So Intentional
Apple unveiled a completely rebuilt "Siri AI" at WWDC 2026 alongside a slew of AI-powered features across Photos, Shortcuts, and Safari — but most of the keynote read like a checklist of things competitors shipped two years ago. To Apple's credit, every demo was conspicuously hands-on and grounded, apparently a direct response to the $250M false advertising settlement over previous AI demos that couldn't actually do what was shown.
Why it matters: Apple's 1.5 billion active devices mean even a cautious, late AI rollout lands with more mass-market reach than almost any competitor.
Microsoft's AI Chief Had to Walk Back His "AI Takes Your Job" Comments Within 24 Hours
Mustafa Suleyman said on the Decoder podcast that AI will automate the jobs of white-collar workers including lawyers, accountants, and project managers — then quickly clarified he meant AI will help those workers complete tasks, not replace them. The backpedal was fast and transparent, which somehow makes it more revealing about where Microsoft's internal thinking actually sits.
Why it matters: When the head of AI at the world's most valuable company can't keep his messaging straight on job displacement, it signals the industry still hasn't figured out how to talk honestly about what's coming.
Lovable Hits $500M ARR With 1 Million New Projects Built Per Week
The AI-powered app builder Lovable announced it's crossed $500 million in annualized run-rate revenue, with users generating a million new projects every week — many of them replacing internal business software entirely. That's not a toy; that's a structural shift in how small teams build and ship software.
Why it matters: Lovable's numbers are the clearest signal yet that AI-native dev tools aren't a niche experiment — they're eating the bottom of the software market at speed.
Meta Is Now Feeding Your Off-Platform Web Activity Into Its AI and Feeds
Meta announced it's expanding how it uses data from third-party websites — games you play, purchases you make elsewhere — to personalize not just ads, but your Facebook and Instagram feeds and AI responses. The program already existed for ad targeting; now it's powering the actual content you see and the AI you talk to.
Why it matters: Folding off-platform behavioral data into AI personalization is a qualitative expansion of surveillance-as-a-service, and most users have no idea it's happening.
Quick Hits
- →Sam Altman's eye-scanning identity startup Tools for Humanity is doing layoffs and reportedly struggling to generate revenue — awkward timing given OpenAI's IPO filing. TechCrunch
- →Apple is waiving AI cloud API costs for indie developers with under 2 million first-time App Store downloads — a smart move to seed the ecosystem before charging for it. TechCrunch
- →73 Microsoft packages were found laced with a credential stealer that self-replicates the moment an AI agent opens them — the second such incident in weeks. Ars Technica
- →Sandstone raised a $30M Series A from Lightspeed and Sequoia to bring AI tooling to in-house corporate legal teams. TechCrunch
AI TLDR