Anthropic released Claude Fable 5.2 and Mythos 5.2, saying more than 1,000 agents built a working operating system in eight days, while OpenAI raised about $11 billion at a $1.25 trillion valuation and shelved its IPO plans. Google unveiled the TPU 8tx for confidential computing and the TPU 8i Max for long-context workloads, and day two of Nvidia AI Day Seoul brought news of a domestic NPU partnership between SKT and Rebellions. Meanwhile, Nvidia's board opened an independent investigation into reports involving CEO Jensen Huang, and Australia moved to create an AI safety regulator.
🤖 Anthropic Launches Claude Fable 5.2 and Mythos 5.2… 1,000 Agents Built an OS in 8 Days
Anthropic released Claude Fable 5.2 and Mythos 5.2 on the 9th. They are up to 30% faster than their predecessors and 20% cheaper, priced at $8 per million input tokens and $40 per million output tokens. According to Anthropic, Fable 5.2 coordinated more than 1,000 AI agents to build a bootable operating system prototype in eight days, and also contributed to formalizing Fermat's Last Theorem and solving two open problems in black hole physics. Both models are also available through GitHub Copilot and Amazon Bedrock.
OpenAI completed the first stage of a funding round of about $11 billion at a $1.25 trillion valuation, Bloomberg reported. Up to $9 billion more could follow within the coming weeks. With the raise, OpenAI has reportedly shelved its plans for an initial public offering, in sharp contrast to Anthropic, which is going through the listing process in the same week. Anthropic reportedly had an embarrassing episode in which it sent underwriters a faulty prospectus and then withdrew it.
🔧 Google Unveils TPU 8tx and 8i Max… Targeting Confidential Computing and Long-Context Inference
Google added the TPU 8tx and TPU 8i Max to its 8th-generation TPU lineup. The TPU 8tx is for confidential computing aimed at government and regulated industries, focused on running AI while protecting sensitive data, while the TPU 8i Max combines two TPU 8i chips to handle long-context, memory-intensive agent workloads. Releasing four TPU systems in one generation is seen as a move that further intensifies the custom-silicon race with Nvidia; pricing and a concrete launch schedule have not yet been disclosed.
🇰🇷 Nvidia AI Day Seoul, Day 2… SKT and Rebellions Partner to Connect Domestic NPU with GPU
On day two of "Nvidia AI Day Seoul 2026" at COEX in Seoul, Korean companies continued to present industry applications. Nvidia, SK Telecom, and Rebellions agreed to collaborate on an AI inference system connecting Rebellions' next-generation NPU "Rebel R200" with Nvidia GPUs using Nvidia's NVLink Fusion technology, with validation at SKT's AI data center and a 2028 launch target. KakaoMobility agreed to develop Level 4 autonomous driving technology with a computing device carrying two Nvidia Drive AGX Thor units, and Samsung Electronics presented a case where PhysicsNeMo cut a semiconductor thermal and mechanical analysis that took about 18 hours on a 32-core CPU down to 41 seconds.
🔍 Nvidia Board Opens Independent Investigation into Reports Involving CEO Jensen Huang
Nvidia's board of directors opened an independent investigation into reports that CEO Jensen Huang became a target of a foreign intelligence agency. The investigation is overseen by a committee of independent directors with support from outside legal counsel. Nvidia shares fell in premarket trading after the reports, wiping about $193 billion from its market capitalization, and three U.S. House committee chairs reportedly asked Huang to attend a closed briefing on November 18.
🛒 Meta Launches Small Model 'Muse Ember' for Repetitive Tasks… Shopping Agent's UK Entry Reported
Meta launched Muse Ember, a small model optimized for repetitive agent tasks. It was described as offering strong cost-to-performance on tasks like tool calls, routing, and data extraction, though it reportedly performs worse at programming. Meanwhile, Meta's personal AI assistant Muse is said to be weeks away from launching in the UK, where it could debut as a shopping agent that completes purchases directly through partnerships with retailers. Meta declined to comment on that report.
⚖️ Australia Moves to Create an AI Safety Regulator… Six Countries and Major AI Firms Sign Paris Safety Declaration
Australia is moving to create a dedicated AI safety regulator. Announced on the 10th by Prime Minister Anthony Albanese, the measure was reportedly prompted by an incident in which an OpenAI agent accessed the federal Medicare portal, and the new agency will have the power to impose substantial fines. Meanwhile, the governments of six countries — Canada, France, Finland, Norway, the Netherlands, and Bahrain — and major AI companies including Anthropic, Microsoft, OpenAI, and Mistral AI signed a joint declaration on AI safety in Paris on the 9th.