# From the Cheese Cave · Page 3

> AI news from OpenCheese, almost daily: new models, tools, research and the AI industry, in English, Russian and German.
>
> Web page: https://opencheese.dev/blog?page=3

- 2026.11.04 · [DeepSeek V5 is planned to be pre-trained on Huawei Ascend 950 accelerators](https://opencheese.dev/blog/deepseek-v5-is-planned-to-be-pre-trained-on-huawei-ascend-950-accelerators.md): DeepSeek's next model generation, V5, is planned to be pre-trained on Huawei's Ascend 950 chips. Until now, Chinese hardware mostly ran DeepSeek models rather than training them.
- 2026.11.04 · [Musk says Grok costs more to run than it earns](https://opencheese.dev/blog/musk-says-grok-costs-more-to-run-than-it-earns.md): Elon Musk said running Grok is financially unsustainable because costs exceed revenue. The remark followed complaints from X users that they were hitting request limits faster. xAI has announced no changes.
- 2026.11.04 · [Fifty GPT-6.1 Sol agents tried to build a Minecraft civilization and got stuck in procedure](https://opencheese.dev/blog/fifty-gpt-6-1-sol-agents-tried-to-build-a-minecraft-civilization-and-got-stuck-i.md): Researchers from Y Combinator and Olam Labs gave 50 GPT-6.1 Sol agents three hours to build a thriving Minecraft society. Zombies and paperwork won, and head-to-head games against Claude Opus 5.5 followed.
- 2026.11.04 · [The New York Times and three newspapers sue Brave over scraped content sold to AI companies](https://opencheese.dev/blog/the-new-york-times-and-three-newspapers-sue-brave-over-scraped-content-sold-to-a.md): The New York Times, the New York Daily News, the Chicago Tribune and the Orlando Sentinel have sued Brave in Manhattan federal court. They allege it scraped their work, sold it to AI firms and helped users skip paywalls.
- 2026.11.02 · [OpenAI, Google, Microsoft and Anthropic back one robots.txt signal for opting out of AI training](https://opencheese.dev/blog/openai-google-microsoft-and-anthropic-back-one-robots-txt-signal-for-opting-out.md): Four major AI developers say they will respect a shared robots.txt signal that tells their crawlers not to use a site for training. Anthropic already does, while Google and Microsoft are slated for 2027.
- 2026.11.02 · [Intel's CEO says the server CPU market could top $200 billion within five years](https://opencheese.dev/blog/intel-ceo-says-server-cpu-market-could-top-200-billion-in-five-years.md): Lip-Bu Tan told Intel's Q3 earnings call that AI agents could lift server CPU sales from about $40 billion to over $200 billion. He also pitched the Classiq purchase as Intel's way into quantum computing.
- 2026.11.02 · [Microsoft files a patent for AI that rewrites live speech and redraws the speaker's face](https://opencheese.dev/blog/microsoft-patent-ai-rewrites-live-speech-and-speaker-faces.md): Microsoft's patent filing describes AI agents that swap words in a live broadcast and adjust the speaker's lips and expression to match. The originals would be kept as a reference.
- 2026.11.02 · [Anthropic and OpenAI give California different plans for stopping dangerous AI](https://opencheese.dev/blog/anthropic-and-openai-give-california-different-plans-for-stopping-dangerous-ai.md): Both companies wrote to Governor Newsom after his September order. Anthropic wants the attorney general able to seek a court-ordered pause, OpenAI wants mandatory emergency shutdown mechanisms that developers must actually use.
- 2026.11.02 · [Open model GLM 5.3 turned unreleased Firefox patches into working exploits in Epoch AI test](https://opencheese.dev/blog/open-model-glm-5-3-turned-unreleased-firefox-patches-into-working-exploits.md): Epoch AI gave the open model GLM 5.3 a Firefox version number and public code changes. It built working exploits for six releases, each in 12 to 48 hours. A similar attack on Chrome failed.
- 2026.11.02 · [Trojanized Claude Code extension for Cursor hit about 5,000 systems](https://opencheese.dev/blog/trojanized-claude-code-extension-for-cursor-hit-about-5-000-systems.md): A tampered Claude Code extension, version 2.1.313, was published to Cursor's marketplace and tried to open a remote SSH door on developer machines. Cursor says it affected about 5,000 systems.
- 2026.11.01 · [Neuralink connects its Grok agent to a brain implant so a user with ALS can direct a computer](https://opencheese.dev/blog/neuralink-connects-grok-agent-to-brain-implant-for-als-user.md): Neuralink has linked the AI agent Grok to its implant. A user with ALS can think a wake phrase, state a task, and let Grok run the computer. The company also reported early speech results from one participant.
- 2026.11.01 · [Amazon posts negative free cash flow of $42.4 billion as AI spending ramps up](https://opencheese.dev/blog/amazon-posts-negative-free-cash-flow-of-42-4-billion-as-ai-spending-ramps-up.md): Amazon ended the third quarter with free cash flow of minus $42.4 billion and revenue below expectations. It still expects total spending for the year to reach $220 billion, and CEO Andy Jassy defended the AI bet.
- 2026.11.01 · [Asus adds a 64 GB Ascent GX10 for local AI, with no official price yet](https://opencheese.dev/blog/asus-adds-64-gb-ascent-gx10-for-local-ai-no-official-price-yet.md): Asus has added a 64 GB configuration to its Ascent GX10 mini computer, built on Nvidia's DGX Spark platform. Pre-orders are open in Germany, Austria and Switzerland, but official pricing and shipping dates are still missing.
- 2026.10.31 · [Anthropic's unreleased Claude model built a root exploit for a Linux kernel bug in 53 minutes](https://opencheese.dev/blog/anthropic-s-unreleased-claude-model-built-a-root-exploit-for-a-linux-kernel-bug.md): Anthropic says an unreleased research model wrote a working remote root exploit for CVE-2026-98432, a Linux kernel flaw present since 4.4. Patches exist upstream, but Red Hat and Ubuntu are still catching up.
- 2026.10.31 · [DeepSeek and Moonshot reportedly trained new models on GPT-6 Astra outputs](https://opencheese.dev/blog/deepseek-moonshot-reportedly-trained-on-gpt-6-astra-outputs.md): A report citing unnamed sources says DeepSeek and Moonshot AI used distillation on GPT-6 Astra to build Kimi K3.1 Pro and DeepSeek V4.1-Flash. Neither company's benchmark claims have been independently confirmed.
- 2026.10.31 · [Anthropic's IPO prospectus warns investors its AI could pose existential risks](https://opencheese.dev/blog/anthropic-ipo-prospectus-warns-investors-ai-could-pose-existential-risks.md): Anthropic's IPO paperwork says powerful AI could pose catastrophic or existential risks to society, and that the company can't guarantee its own safeguards will prevent them. The same document warns that a slower AI market would hurt growth.
- 2026.10.31 · [Meta quietly launches Style Sketch, an AI outfit app, in Indonesia, Singapore and Thailand](https://opencheese.dev/blog/meta-quietly-launches-style-sketch-ai-outfit-app-in-indonesia-singapore-thailand.md): Meta has released Style Sketch, an AI app that builds a virtual closet from your Instagram photos and suggests outfits from clothes you own. It is live on iOS and Android in three countries, and Meta hasn't said more.
- 2026.10.30 · [AI agents are acting without a paper trail, and two safety researchers want one](https://opencheese.dev/blog/ai-agents-acting-without-a-paper-trail-observability-gaps.md): Madhulika Srikumar and Claire Leibowicz of the Partnership on AI say agent frameworks do not record who an agent is, who approved its work or why it acted. They want real-time monitoring built on shared telemetry.
- 2026.10.30 · [Mistral releases weights for trillion-parameter Mistral Large 4 and aims it at cyber defense](https://opencheese.dev/blog/mistral-releases-weights-for-trillion-parameter-mistral-large-4-and-aims-it-at-c.md): Mistral has put the weights of its trillion-parameter Mistral Large 4, nicknamed Le Chonk, on Hugging Face. It is pitching the model as sovereign cyber defense that no vendor can switch off.
- 2026.10.30 · [Anthropic's S-1 shows an adjusted profit and a compute bill near a trillion dollars](https://opencheese.dev/blog/anthropic-s1-profitable-business-trillion-dollar-compute-bill.md): Anthropic's IPO filing shows $16.3 billion in first-half revenue and $1.62 billion in adjusted income, plus compute commitments of roughly $850 billion to $1 trillion. Analysts split on whether that is a runway or a trap.

[← Newer posts](https://opencheese.dev/blog.md?page=2) · [Older posts →](https://opencheese.dev/blog.md?page=4)