Evening Scan

Happy Fourth Eve. The day's big story is what didn't happen: AI agents remain stubbornly non-functional, Meta's internal model timeline is slipping, and Anthropic is preparing for an IPO while cutting off Chinese customers. Markets are closed tomorrow, so today was mostly administrative.

Mark Zuckerberg Tells Staff AI Agents Haven't Progressed as Quickly as He'd Hoped

At a town hall, Zuckerberg said Meta's agentic AI reorganization wasn't as "clean" as he'd hoped — a public wince from a CEO. (TechCrunch)

Anthropic's Bankers Hire Freshfields to Advise on IPO

The UK firm that advised on Google's Wiz acquisition is now advising Anthropic — IPO preparations are underway. (The Information)

Anthropic Moves to Close Loopholes Letting Chinese Firms Access Its Models (paywall)

Ant Group and others had been routing around export controls via cloud providers and overseas subsidiaries. Anthropic is now actively closing those gaps ahead of its IPO. (Financial Times)

Tesla Plans $200-Per-Week Cap on Employee AI Spending Starting July 6

The limit excludes beta xAI products, nudging staff toward the in-house option. (The Information)

“Sharper, Smarter, More Cunning, Brutal” — Flipper CEO describing the Flipper One. Truly the only honest product launch quote of 2026.