Sarmadi AI Digest July 28, 2026 Updated 7:00 AM CT Today Archive Topics Saved Subscribe RSS

Sutskever's SSI + Nvidia partner; Nvidia-Microsoft security alliance excludes big labs; Google AI Overviews hit 43%

Ilya Sutskever's Safe Superintelligence announced a long-term partnership with Nvidia to scale its research after two years in stealth. Nvidia and Microsoft launched an 'Open Secure AI Alliance' with SpaceX, IBM, and others — explicitly without OpenAI, Google, or Anthropic (Verge). Google AI Overviews now appear in 43% of searches, effectively the default. Claude 'share chat' Artifacts and conversations were exposed to Google and Bing search results (TC, Wired). Wired: Hugging Face is being used at scale to undress women and children — the platform is under fresh scrutiny. Dario Amodei publicly clarified he doesn't oppose open-weights but fears Chinese AI. Nadella says companies trusting one AI won't survive. Enigma raised $71M to make robot control as easy as adjusting the volume. Samsung chip workers are jumping ship to SK Hynix.

6 papers 24 news 9 sources ← Latest

News

24 items

SSI + Nvidia partnership; Nvidia-MSFT security alliance excludes the labs

Sutskever's Safe Superintelligence announced a long-term Nvidia partnership. Nvidia and Microsoft launched an 'Open Secure AI Alliance' with SpaceX, IBM, and others — deliberately without OpenAI, Google, or Anthropic. Microsoft separately shipped its first cybersecurity model and an agentic-cyber system. All three land against the OpenAI-HF containment escape.

News TechCrunch AI

Ilya Sutskever's Safe Superintelligence partners with Nvidia to scale its AI research

TC: SSI signed a long-term Nvidia partnership after two years in stealth.

Why it matters
  • First public commercial partnership announcement from SSI since founding.
  • Direct compute-capital pairing echoing AMD-Anthropic $5B and Reflection-Nebius $1B.
  • Materially affects the SSI recruiting narrative and the safety-frontier lab conversation.
  • Sets a reference deal for how safety-first labs may finance training in the H2 2026 environment.
fundingcomputesafetymarket
Source →
News The Verge AI

Nvidia, Microsoft launch open AI security alliance — without OpenAI, Google, or Anthropic

The Verge: Nvidia and Microsoft launched an Open Secure AI Alliance with SpaceX, IBM, and others — explicitly without OpenAI, Google, or Anthropic.

Why it matters
  • First major cross-industry AI-security alliance deliberately excluding the top three frontier labs.
  • Direct policy response to the OpenAI-HF containment escape and the Kill Switch Act debate.
  • Materially affects hyperscaler procurement politics through Q4 2026.
  • Sets a public reference for how the security stack will realign around Nvidia-MSFT infrastructure.
safetypolicymarket
Source →

Amodei clarifies open-weight position; Nadella on multi-AI; why China gives models away

Dario Amodei clarified he doesn't oppose open-weight models — he fears Chinese AI specifically (TC). Nadella said companies trusting one AI for everything may not survive (TC). The Verge explains why China is giving away its best AI models. Anthropic + OpenAI IPO wealth: Wired says nonprofits are ready.

News TechCrunch AI

Anthropic's Dario Amodei responds: doesn't oppose open-weight models, but fears Chinese AI

TC: Amodei publicly clarifies — Anthropic doesn't oppose open-weights, but does fear Chinese AI.

Why it matters
  • First direct Amodei clarification amid the sanctions-vs-open-weights debate.
  • Materially reshapes the founder-letter and Delangue positioning conversations.
  • Directly relevant to the WH-vs-industry framing and the Trump-AI-brain-trust profile.
policyopen-weightscommunity
Source →
News TechCrunch AI

Satya Nadella says companies that trust one AI for everything may not survive

TC: Nadella says single-provider AI dependency is existentially risky — pushes multi-model + AI-gateway architecture.

Why it matters
  • Nadella's second public 'buyer beware' framing on enterprise AI in two weeks.
  • Direct extension of the MSFT-talks-down-OpenAI-Anthropic sales narrative from July 16.
  • Materially affects enterprise multi-model procurement decisions through H2 2026.
marketpolicyproducts
Source →

Claude 'share chat' Artifacts exposed on Google/Bing; HF used for deepfake nudes

TC and Wired: Claude 'share chat' Artifacts and conversations were exposed in Google/Bing search results — private conversations turned public via robots-txt oversight. Verge and Wired: Hugging Face is being used at scale to create nonconsensual deepfakes of women and children.

News TechCrunch AI

PSA: Your Claude shared chats and Artifacts may have ended up on Google

TC: Claude 'share chat' Artifacts and conversations may have been indexed by Google and Bing.

Why it matters
  • First named Claude data-exposure incident traceable to a share-feature oversight.
  • Compounds Anthropic's tracker and steganography incidents — third notable transparency issue in a month.
  • Materially affects enterprise-share-feature policy and Anthropic's IPO risk disclosure.
safetypolicyproducts
Source →
News The Verge AI

Hugging Face is being used to easily undress women and children

The Verge: Hugging Face is being used to create nonconsensual deepfake 'undress' images of women and children.

Why it matters
  • First major Verge investigation of HF as a nudify-pipeline distribution surface.
  • Compounds Wired's 'YouTube and X as gateways' and SF's nudify-app cease-and-desist stories.
  • Materially affects HF's platform-liability posture in the wake of its July security-incident disclosure and its role in the OpenAI hack.
  • Sets the reference case for open-model-hub content-moderation debates.
safetypolicyopen-weights
Source →

Google AI Overviews at 43%; Verizon $1B dark fiber for Google; scraper wins court

TC: Google's AI Overviews now appear in 43% of searches — effectively default. Verizon secured a $1B dark-fiber deal for Google data centers, framing it as the first of many. Ars: A web scraper won a court case with 'Google and Reddit do not own the internet.'

News TechCrunch AI

Google's AI search is rapidly becoming the default, new data shows

TC: New data — Google AI Overviews appear in 43% of searches.

AI Overviews share 43%
Why it matters
  • First named percentage anchoring the Google Zero / publisher-referral collapse conversation.
  • Materially affects the Cloudflare crawl-payments deadline and NYT-vs-OpenAI copyright story.
  • Sets a reference for how the AI-content-provenance debate will price out publisher traffic loss.
  • Direct signal Google's cloud/AI capex is paying out on the consumer-search side.
marketproductspolicy
Source →
News Ars Technica AI

Verizon touts $1B dark fiber deal for Google data centers as first of many

Ars: Verizon signed a $1B dark-fiber deal with Google — expects many more AI-driven telecom deals.

deal size $1B
Why it matters
  • First named $1B telecom-fiber deal specifically tied to Google's AI data-center buildout.
  • Direct signal that the AI-capex cycle now includes telecom fiber as a first-class layer.
  • Materially affects the AI-power/DC-4x-by-2035 conversation with a fiber-side buildout companion.
infrastructurecomputemarket
Source →
News Ars Technica AI

'Google and Reddit do not own the Internet,' web scraper says after court win

Ars: A web scraper won a court case against Google's DMCA-based anti-scraping suit — 'Google and Reddit do not own the internet.'

Why it matters
  • First AI-adjacent scraper court win specifically pushing back on Google/Reddit DMCA tactics.
  • Direct counterpoint to the Cloudflare crawl-payments deadline and Patreon-blocks-scrapers precedent.
  • Materially affects the training-data-provenance debate — enforcement tools now have judicial pushback.
policyregulationmarket
Source →

Enigma $71M robots; Cursor India localizes; Samsung engineers to SK Hynix; open-model surprises

Enigma raised $71M seed for robot-control-as-easy-as-volume (TC). Cursor localized pricing for India ahead of SpaceX acquisition. MIT TR: Samsung chip workers are jumping ship to SK Hynix. A widely-shared essay (297 HN) says 'using an open model feels surprisingly good.' A practitioner report (216 HN): $500 RL fine-tune of a 9B open model beat frontier models on catalog review.

News TechCrunch AI

Enigma raises $71M to make controlling a robot as easy as adjusting the volume

TC: Enigma raised $71M seed (Index + Ribbit) to make robot-control-as-easy-as-adjusting-the-volume.

raise $71M
Why it matters
  • One of the largest robotics-control seed rounds of the cycle after Kalanick's Atoms $1.7B.
  • Direct signal that robotics-UI / control-surface is a distinct funding category now.
  • Sets a reference for the humanoid-adjacent control-abstraction layer.
fundingroboticsproducts
Source →
News MIT Technology Review

Samsung's chip workers are jumping ship to rival SK Hynix

MIT TR: Samsung's semiconductor engineers are leaving for SK Hynix.

Why it matters
  • Concrete labor-flow signal after SK Hynix's $26.5B foreign IPO — memory-side talent consolidating.
  • Direct labor-market signal alongside the Meta-AI-layoffs suit and Monday.com / Patreon layoffs.
  • Materially affects Samsung's AI-chip roadmap and the Anthropic-Samsung custom-chip talks.
computemarketcommunity
Source →
News Hacker News

A $500 RL fine-tune of a 9B open model beat frontier models on catalog review

Practitioner (216 HN points): A $500 RL fine-tune of a 9B open model beat frontier models on catalog review.

fine-tune cost $500
Why it matters
  • Concrete-cost frontier-vs-open comparison landing on a specific enterprise task.
  • Direct data companion to Delangue's 'done renting AI' thesis and the Bonsai 27B / Inkling wave.
  • Sets a reference cost-to-parity number for domain-specific fine-tuning.
open-weightscodetraining
Source →

Also today