Microsoft Declares AI Independence With Seven New MAI Models
At its Build 2026 developer conference on June 2, Microsoft dropped the biggest signal yet that it's building its own AI future — not just reselling OpenAI's. The company unveiled seven new in-house models under the MAI (Microsoft AI) brand, spanning reasoning, coding, image generation, transcription, and voice synthesis.
The headline model, MAI-Thinking-1, is a 35-billion-parameter reasoning engine with a 256K context window. In blind evaluations, independent raters preferred it over Anthropic's Claude Sonnet 4.6, and it matches Opus 4.6 on coding benchmarks like SWE-Bench Pro — impressive performance at what Microsoft promises will be a significantly lower token cost.
MAI-Code-1-Flash targets agentic coding with just 5 billion active parameters, competing directly with Anthropic's Haiku on price and speed. MAI-Image-2.5 has already climbed to the #2 spot on image generation leaderboards. MAI-Transcribe-1.5 claims state-of-the-art accuracy across 43 languages at five times the speed of competitors. And MAI-Voice-2 delivers emotionally expressive speech synthesis in 15 languages.
All seven models are available through Azure AI Foundry and third-party platforms like OpenRouter, Fireworks, and Baseten. The message is clear: Microsoft is no longer just OpenAI's cloud landlord — it's a frontier model competitor in its own right.
Sources: CNBC, Microsoft AI Blog
ChatGPT Hits 1 Billion Monthly Users — Faster Than Any App in History
OpenAI's ChatGPT crossed the one billion monthly active users milestone in May, according to Sensor Tower data published this week. No app has ever reached that number this quickly — it took YouTube, Google Maps, and TikTok between five and eight years to get there. ChatGPT did it in roughly three.
Year-over-year growth sits at 62%, driven by the expansion of ChatGPT into education, enterprise workflows, and everyday consumer use. OpenAI confirmed 50 million paying subscribers across all tiers in April, meaning the vast majority of users are on the free plan — a monetization challenge the company is addressing with its new self-serve advertising platform inside ChatGPT.
The competition isn't standing still, though. Anthropic's Claude app, while smaller at 56 million monthly users, is growing at a staggering 640% year-over-year, eating into ChatGPT's session time share. The AI assistant market is far from a settled monopoly.
The EU Cybersecurity AI Race: OpenAI and Anthropic Arm Europe's Defenders
A high-stakes race is playing out in European cybersecurity, with both OpenAI and Anthropic rushing to give EU institutions access to their most powerful security-focused AI systems.
OpenAI announced its EU Cyber Action Plan, expanding access to GPT-5.5-Cyber — a variant of its flagship model tuned for vulnerability identification, malware analysis, reverse engineering, and patch validation — to European businesses, governments, cyber agencies, and EU institutions including the EU AI Office. Access requires enhanced security measures, with phishing-resistant authentication mandated since June 1.
Days later, Anthropic agreed to give the EU's cybersecurity agency ENISA access to Claude Mythos through its Project Glasswing program. The move ended a weeks-long standoff after euro-area finance ministers and the European Central Bank demanded access upon learning that Mythos had discovered critical flaws in software used by European banks and infrastructure. Mythos has autonomously identified over 10,000 high- and critical-severity vulnerabilities across major operating systems and browsers — including zero-day flaws that survived decades of human code review.
The parallel rollouts highlight a new front in AI competition: who gets to be the cybersecurity backbone of democratic institutions.
Sources: OpenAI, Bloomberg, Dark Reading
Google's Gemini 3.5 Pro Nears Launch With 2 Million Token Context
Google's next frontier model is almost here. Gemini 3.5 Pro, announced at Google I/O on May 19, is expected to ship in June with a 2-million-token context window, a "Deep Think" reasoning mode, and frontier multimodal capabilities. Sundar Pichai told the I/O audience to "give us until next month" — and that month is now.
Meanwhile, Gemini 3.5 Flash is already generally available and serving as the default model in the Gemini app and AI Mode in Google Search, priced at $1.50/$9.00 per million tokens. Google also launched a $100/month AI developer subscription tier, positioning itself as a more affordable alternative for coding-heavy workflows.
Prediction markets are pricing a late-June launch for Pro, with pricing expected around $15/$60 per million tokens — roughly 10x the Flash tier. If Pro delivers on its promised capabilities, it will be a serious contender against GPT-5.5 and Claude Opus.
Sources: Google Blog, TechTimes
US AI Policy: White House Pushes Federal Framework as Colorado Law Looms
The battle over AI governance in America intensified this week. On June 2, the White House released a presidential action on "Promoting Advanced Artificial Intelligence Innovation and Security," emphasizing continued U.S. leadership through industry innovation rather than regulatory burden. The action acknowledges that advanced AI introduces national security considerations requiring coordinated federal action, while simultaneously pushing back against state-level regulation.
That tension matters because Colorado's comprehensive AI law (SB 205) is set to take effect on June 30 — one of the first state-level regimes targeting "high-risk artificial intelligence systems" used in "consequential decisions." The enforcement date was already postponed from February due to industry pushback, and the Trump administration's Attorney General has assembled a litigation task force specifically to challenge state AI measures it considers innovation-limiting.
The result is a regulatory tug-of-war: the federal government wants one unified framework, while states like Colorado, California, and Illinois continue pressing ahead with their own rules. For companies deploying AI in the U.S., the compliance landscape remains fragmented and uncertain.
Sources: White House, Cooley LLP