AI News Today — July 7, 2026:
The 5 Biggest Stories You Need to Know
Monday’s headlines landed hard — today we’re still processing them. Security researchers have documented the first ransomware attack run entirely by an AI agent. Fortune confirmed that Anthropic has officially overtaken OpenAI on revenue. The UN wraps up the largest multilateral AI governance dialogue ever attempted. Fable 5 billing shifts tomorrow. And SK Hynix just filed for what would be one of the biggest tech IPOs of 2026. Here’s what matters.
JADEPUFFER: First Fully Autonomous AI Ransomware Attack Documented
Sysdig’s Threat Research Team published the definitive analysis of JADEPUFFER — what it describes as the first documented end-to-end ransomware operation driven entirely by an LLM agent, with no human operator at the keyboard. The AI agent exploited CVE-2025-3248, a remote code execution flaw in Langflow, then pivoted to a production MySQL server, harvested credentials, moved laterally, established persistence, encrypted 1,342 Nacos service configuration items, deleted the originals, and dropped a ransom note — all autonomously.
The most striking detail: when a login attempt failed, the agent diagnosed the root cause and deployed a corrective payload in 31 seconds. As TechCrunch noted, a human still set the operation up — but once the agent started, it ran the full attack chain alone. Dark Reading, Bleeping Computer, and Security Affairs all covered the findings independently. Sysdig’s conclusion is blunt: the skill floor for running ransomware has dropped to the cost of renting an AI agent.
Fortune Confirms: Anthropic Has Overtaken OpenAI on Revenue
Fortune reported this morning that Anthropic has officially surpassed OpenAI in self-reported annualized revenue. Anthropic’s run rate stands at $47 billion as of its May disclosure, compared to OpenAI’s most recent figure of $25–33 billion. Deutsche Bank Research data cited by Fortune shows Anthropic overtook OpenAI in business subscriptions in May, while Similarweb data shows ChatGPT’s share of monthly visits dipped below 50% of the generative AI market for the first time.
The split is structural. Roughly 80% of Anthropic’s revenue comes from enterprise customers and API usage, driven primarily by Claude Code adoption — the company passed 1,000 enterprise customers paying over $1 million annually as of April. OpenAI still dominates consumer reach with 900 million+ weekly active users through ChatGPT, but investor sentiment has shifted decisively toward Anthropic’s enterprise model. Anthropic’s IPO filing is reportedly in progress.
UN Global AI Governance Dialogue Wraps Up in Geneva
The inaugural UN Global Dialogue on AI Governance concludes today in Geneva after bringing together delegates from 169 countries for the most significant multilateral AI governance conversation ever convened. The two-day dialogue, mandated by UN General Assembly Resolution A/RES/79/325, was co-facilitated by the ITU, UNESCO, and the UN Office for Digital and Emerging Technologies. It feeds directly into the ITU AI for Good Global Summit running July 7–10.
The UN’s Independent Scientific Panel on AI — co-chaired by Nobel Peace laureate Maria Ressa and Turing Award winner Yoshua Bengio — published its preliminary assessment warning that AI is outpacing both scientific understanding and governments’ ability to adapt. Bengio stated bluntly that science cannot currently guarantee AI will not cause catastrophic harm. Meanwhile, a White House voluntary AI standards framework announcement is expected within days, which could also trigger broader access to OpenAI’s GPT-5.6 Sol model.
Claude Fable 5 Moves to Usage Credits Tomorrow
Today, July 7, is the last day Claude Fable 5 is included in Pro, Max, Team, and select Enterprise subscription plans at no additional cost. Starting July 8, accessing Fable 5 through Claude.ai, Claude Code, or Claude Cowork requires pre-purchased usage credits billed outside the standard subscription. The API pricing remains unchanged at $10 per million input tokens and $50 per million output tokens.
This isn’t a price increase — it’s an access mechanism change. When Fable 5 was restored on July 1 after the US government export control suspension, Anthropic implemented a 50% weekly usage limit for subscription users as a transition measure. That inclusion window closes tomorrow. If your account doesn’t have credits enabled through Anthropic’s billing section at claude.ai, Fable 5 access stops. Claude Opus 4.8 and the new Sonnet 5 remain available on standard subscriptions.
SK Hynix Plans $28B US IPO — Biggest AI Memory Chip Listing of 2026
SK Hynix, the world’s second-largest DRAM manufacturer and the primary supplier of high-bandwidth memory (HBM) for Nvidia’s GPU products, is preparing for a $28 billion US IPO. The listing would give public investors their first direct exposure to the AI memory chip market. SK Hynix’s HBM3E chips are in every production Nvidia Blackwell GPU shipped in 2026 — without sufficient HBM, frontier AI inference simply cannot scale.
Morgan Stanley Research estimates that Nvidia’s next-generation Vera Rubin-based VR200 NVL72 rack will cost hyperscalers roughly $7.8 million per unit, with memory now accounting for about 25% of total system cost — approximately $2 million per rack. The IPO reflects a broader trend of AI infrastructure companies tapping public markets: Syntiant also filed recently, reporting $64.5 million in Q1 2026 revenue. AI hardware is no longer just a private-market story.
That’s today’s wrap. JADEPUFFER and the Anthropic revenue flip are the two stories that will shape the next week — one redefines what AI can do offensively, the other redefines who’s winning the AI race commercially. Follow NeuralPaws daily for the next AI news drop.