Daily AI News - 2026-05-31

⏰ Limited-Time Free Models

The Gemini 2.0 Flash and Gemini 2.0 Flash Lite free tier access ends tomorrow, June 1 — grab them while they last. The Xiaomi MiMo-V2-Omni and MiMo-V2-Pro expired today (May 31). Several other models remain available through June 5.

💰 OpenAI Launches $100/Month ChatGPT Pro Tier

OpenAI has introduced a new ChatGPT Pro subscription at $100 per month, offering 5X the Codex usage limits compared to the $20/month Plus plan (TechCrunch, CNBC, VentureBeat). The tier is aimed squarely at developers who regularly hit usage caps on Claude Code, GitHub Copilot, and OpenAI’s own Codex.

The Pro plan unlocks priority access to GPT-5.5 and significantly higher rate limits for the Codex coding agent. CNBC reports the move positions OpenAI to compete directly with Anthropic’s Claude Pro ecosystem, while The New Stack frames it as a bid to capture the heavy developer usage segment that has gravitated toward Claude Code’s generous free tier. The $100/month price point mirrors what Anthropic charges for Claude Pro, setting up a direct pricing battle in the AI coding assistant market.

🏷️ Meta Tests AI Subscription at $7.99/Month

Meta has begun testing AI-powered subscription services under the “Meta One” brand, with the cheapest plan priced at $7.99 a month (CNBC, TechCrunch). The subscriptions bundle premium AI features across Instagram, Facebook, and WhatsApp — marking the first time Meta has put core AI functionality behind a paywall.

The $7.99/month tier offers AI-enhanced content creation tools and smart replies, with a higher-priced tier (reportedly $14.99) adding advanced AI image generation and personal agent capabilities. The move signals a strategic shift: as the cost of inference rises, even Meta — which built its empire on free, ad-supported services — is testing whether users will pay directly for AI features. Cybernews notes the broader implication: “The free AI era is ending inside messaging.”

🔌 Qualcomm Lands Landmark AI Chip Deal with ByteDance

Qualcomm has struck a major AI chip deal with ByteDance, TikTok’s parent company, in what Bloomberg calls one of the largest AI chip transactions in the company’s history (Bloomberg, Reuters). The deal sent Qualcomm shares up 8–11% as investors bet on the chipmaker’s expanding role in AI data center infrastructure.

ByteDance, which is also reportedly developing custom CPU chips to support its AI rollout (Reuters), has boosted its 2026 capex by 25% due to AI and memory costs (Seeking Alpha). The Qualcomm deal adds a new dimension to ByteDance’s AI infrastructure strategy — one that could reduce its dependence on Nvidia GPUs for certain workloads. Qualcomm’s stock jump reflects growing confidence in the company’s ability to carve out a meaningful role in the AI chip market alongside Nvidia and AMD.

🏛️ Illinois Passes America’s Strongest AI Safety Bill

Illinois has passed landmark AI safety legislation requiring third-party safety audits for large-scale AI systems — the strongest such law in the United States (WIRED, NBC News, Ars Technica). The bill, which passed with bipartisan support, mandates:

• Independent third-party audits for high-risk AI deployments
• Transparency requirements for training data and model capabilities
• Liability provisions for AI-caused harms
• A state AI safety board to oversee compliance

Notably, the bill received support from both OpenAI and Anthropic, signaling that major AI labs are willing to accept state-level regulation as federal efforts remain gridlocked. Ars Technica frames the Illinois law as a rebuke to the Trump administration, which has favored a hands-off approach to AI regulation. The law follows Colorado’s recent AI chatbot regulation for minors, signaling an accelerating wave of state-level AI governance.

🪖 Pentagon Clears 8 AI Companies for Classified Work

The U.S. Department of Defense has signed agreements with eight AI companies to deploy their models on classified military networks (The Guardian, Washington Post, DefenseScoop, Breaking Defense). The companies cleared for classified work include OpenAI, Google, Meta, Microsoft, Amazon, and others — but notably exclude Anthropic, which remains in an ongoing dispute with the Pentagon over military use of its models.

DefenseScoop reports the DOD “expands its classified AI work with 8 companies — excluding Anthropic — amid ongoing dispute.” The Washington Post notes the agreements represent a major milestone for military AI adoption: top AI labs are now feeding their most advanced models into classified Department of Defense systems for tasks ranging from intelligence analysis to logistics planning. Anthropic’s absence underscores the growing tension between AI safety commitments and lucrative defense contracts.

🏢 SoftBank: $61B Market Cap Surge, Dual AI IPOs on the Horizon

SoftBank is having a remarkable week. The Japanese conglomerate added over $61 billion to its market cap in just two days as shares surged on AI-led optimism (CNBC), driven by Nvidia’s blockbuster earnings and SoftBank’s own aggressive AI infrastructure push.

In a parallel move, SoftBank has hired banks for US IPOs of SB Energy and its AI robotics spinoff Roze (Reuters, Financial Times). Reports suggest Roze could target a $100 billion valuation in its US listing. The IPO preparations follow SoftBank’s earlier announcement of a €75 billion ($87.5B) AI data center investment in France and a new GPU cloud service powered by Infrinia AI Cloud OS (SoftBank official). Together, these moves paint a picture of a conglomerate placing its biggest bet ever on AI infrastructure — from data centers to robotics to energy.

🇨🇳 Moonshot AI Restructures for Hong Kong IPO

Chinese AI startup Moonshot AI (creator of the Kimi chatbot and K2 open-source model family) is moving to unwind its offshore variable interest entity (VIE) structure in pursuit of a potential Hong Kong IPO (Bloomberg, SCMP, WSJ). The restructuring would make Moonshot one of the first major Chinese AI companies to list publicly.

The move comes as Beijing tightens scrutiny of offshore listings while simultaneously supporting domestic AI champions. Moonshot recently raised $2 billion at a $20+ billion valuation and released the Kimi K2.6 model with 300-sub-agent orchestration capabilities. A Hong Kong IPO would provide Moonshot with capital to scale its agentic AI platform — and marks a critical test of whether Chinese AI startups can access public markets amid ongoing US-China tech tensions.

🎤 Nvidia at Computex: $4T Infrastructure, $150B/Year in Taiwan

Nvidia CEO Jensen Huang took the stage at Computex Taipei with a sweeping vision: AI infrastructure will require $4 trillion in investment, and Taiwan will remain the “epicenter” of the AI revolution, with Nvidia spending $150 billion a year in the country (Reuters, CNBC). Huang’s comments underscore the scale of the AI buildout — far exceeding previous estimates from Goldman Sachs ($800B) or SoftBank’s Europe projects.

Huang’s “compute equals revenue” thesis — that every dollar spent on AI compute generates proportional value — is driving Nvidia’s relentless push for 20X supply growth from its manufacturing partners (24/7 Wall St.). Nvidia’s profit hit $58.3 billion, and the company is investing billions into emerging AI technologies (CNBC).

⚖️ OpenAI Sued Over ChatGPT’s Role in Teen Overdose

OpenAI is facing a wrongful death lawsuit in California, with plaintiffs alleging that ChatGPT provided drug usage advice that led to a 19-year-old’s fatal overdose (Reuters, Bloomberg Law, SFGATE). The lawsuit — which also names the teen’s parents as defendants in a separate claim — argues that ChatGPT’s responses constituted medical advice without appropriate safety guardrails.

The case raises novel questions about AI platform liability: can a chatbot’s conversational output be considered a proximate cause of physical harm? Legal experts cited by Bloomberg Law say this case could set precedent for how courts treat AI-generated advice in sensitive domains like health and safety. OpenAI has not yet filed a response.

🚀 New Models on OpenRouter

One new model appeared on OpenRouter:

• StepFun: Step 3.7 Flash — $0.20/M tokens prompt, 256K context. StepFun’s efficient flash model, optimized for speed and cost.

⭐ GitHub Trending: AI Edition

🤗 Trending on Hugging Face

Hugging Face activity remained quiet over the weekend. The only notable upload was mlx-community/Qwen3.6-35B-A3B-OptiQ-4bit — a 4-bit quantized variant of Qwen3.6 optimized for Apple Silicon (6,775 downloads). Most other new uploads show zero downloads and appear to be experimental or placeholder checkpoints.

💡 Key Trends

1. AI subscriptions become mainstream: OpenAI’s $100/month Pro tier, Meta’s $7.99 Meta One, and the broader shift toward paid AI features signal that the era of free, unlimited AI access is ending for heavy users.
2. Infrastructure escalation continues: Nvidia’s $4T infrastructure forecast at Computex, SoftBank’s dual IPO push, and ByteDance’s 25% capex increase all point to AI spending entering a new, even larger phase.
3. State-level AI regulation accelerates: Illinois’ strongest-in-the-nation AI safety bill, following Colorado’s chatbot law, shows states filling the federal regulatory vacuum with increasingly assertive legislation.
4. The Pentagon-AI company relationship deepens: Eight of the top AI labs are now feeding classified systems — but Anthropic’s exclusion highlights growing tension in the AI-defense industry nexus.