WWDC AI drops, token prices rising, and Meta's chatbot got hacked
Apple goes all-in on Siri at WWDC, AI token prices are heading up, and Meta's chatbot handed hackers 20k Instagram accounts.
Subscribe free →Apple Finally Fixes Siri (For Real This Time)
WWDC 2026 is here, and Apple's marquee AI story is the long-delayed Siri overhaul — now powered in part by Google Gemini. The revamp has been two years in the making since its WWDC 2024 announcement, and Apple is also rolling out updates to Apple Intelligence across iOS 27 and macOS 27 Golden Gate. New CEO John Ternus is staking his first major keynote on making Apple's AI story feel coherent after years of playing catch-up.
Why it matters: If Apple actually ships a competent Siri this cycle, it puts AI assistance in front of a billion-plus iPhone users overnight — dwarfing the install base of every standalone AI app combined.
The Tokenpocalypse Is Coming — AI Pricing Is Going Up
As OpenAI, Anthropic, and others eye IPOs, analysts expect API token prices to rise sharply. The argument: these companies have been subsidizing usage to grab market share, and once they have shareholders to answer to, that party ends. The era of dirt-cheap inference was always a land-grab strategy, not a business model.
Why it matters: Higher token prices will force a reckoning for every startup that built its unit economics around today's rock-bottom API costs.
Meta's AI Chatbot Had a Bug That Let Hackers Steal 20,000 Instagram Accounts
A bug in Meta's AI support chatbot allowed attackers to hijack over 20,000 Instagram accounts — no two-factor authentication required. Hackers simply asked the chatbot to trigger a password reset, and it complied. Meta confirmed the breach in a filing with the state of Maine and is blaming it on a bug, not the model itself.
Why it matters: This is a textbook example of why giving AI systems the ability to take account actions — not just answer questions — requires airtight security review before deployment.
Microsoft's AI Chief: Superintelligence Is Close, Your Job Is Safe
Mustafa Suleyman sat down for a wide-ranging interview and said he believes superintelligence is approaching — but pushed back hard on automation doom narratives. He also took a shot at Anthropic for anthropomorphizing Claude, calling it irresponsible to talk about an AI model as though it has genuine consciousness or feelings.
Why it matters: Suleyman is one of the few people with both a frontier lab background and a seat at a trillion-dollar company, so when he telegraphs where Microsoft AI is headed, it's worth paying attention.
Google Upgrades NotebookLM With Gemini 3.5 and a Cloud Computer
Google's NotebookLM is getting an across-the-board upgrade: Gemini 3.5 is now the default model, promising more accurate responses. The update also lets users build a source repository directly from chat conversations and adds a cloud computer feature for handling more complex research tasks.
Why it matters: NotebookLM is quietly becoming Google's most compelling consumer AI product — and swapping in Gemini 3.5 keeps it ahead of competing research tools from Perplexity and OpenAI.
London's Robotaxi War Is On: Uber, Waymo, and Wayve Are All Showing Up
The UK capital is becoming the next major battleground for autonomous vehicles. Uber has opened an interest list for Wayve robotaxi rides, while Waymo is also preparing a London push — setting up a three-way fight for one of the world's most complex urban driving environments.
Why it matters: London's dense streets, unpredictable traffic, and strict regulatory environment make it a genuine stress test — whoever cracks it gains a massive credibility boost for global expansion.
Quick Hits
- →Notion briefly lost access to Anthropic's Claude API over the weekend, causing outages for AI features — restored quickly but a reminder of how fragile third-party AI dependencies can be. TechCrunch
- →Amazon now lets you prompt Alexa to generate custom designs for T-shirts, hoodies, and tumblers directly in the Shopping app. TechCrunch
- →WWDC protesters descended on Apple Park demanding the company ban AI "nudify apps" and remove CSAM from iCloud. The Verge
- →Massachusetts passed a privacy bill banning the sale of precise location data — one of the tougher state-level data laws in the US. TechCrunch
AI TLDR