Today’s briefing covers a widening scandal at Nvidia that has triggered congressional export crackdowns and a security leadership shakeup, alongside Anthropic’s release of Claude Fable 5.2, a model that generates images by simulating brushstrokes through code rather than diffusion.
Key takeaways
- Nvidia faces bipartisan congressional pressure and export restrictions after revelations that a partner linked to its former CEO funneled billions in controlled chips to Chinese firms.
- Nvidia’s chief security officer and product security head departed in August 2026 amid the Megaspeed investigation and insider threat concerns.
- Anthropic’s Claude Fable 5.2 introduces a code-driven, physics-based painting approach that enables step-by-step planning and self-correction, outperforming rivals on robotics benchmarks.
Nvidia CEO scandal export crackdown security shakeup
Documents reviewed by the Wall Street Journal show that Megaspeed arranged nearly $12 billion in Nvidia high-performance chip orders for Chinese companies between March 2024 and June 2025, covering several hundred thousand chips the US sought to block via export controls, according to The Decoder. Megaspeed is owned by Alice Huang, the former partner of ousted Nvidia CEO Jensen Huang, and gained preferential access as an Nvidia cloud partner partly due to its relationship with Alibaba, The Decoder reports. Nvidia officially reported $12.4 billion in China revenue over six months, or 7 percent of total revenue, while short-seller Culper Research estimates Chinese customers accounted for more than 20 percent of last fiscal year’s revenue, per The Decoder.
The New York Times reported in October 2025 that Megaspeed was under investigation for helping Chinese customers obtain export-controlled chips, The Decoder notes. Following revelations about Jensen Huang’s relationship with a suspected Chinese spy, Congress is moving to further restrict AI chip exports, with both parties supporting tighter rules and demanding testimony from Huang and Commerce Secretary Howard Lutnick, according to The Verge. Senate Minority Leader Chuck Schumer and Banking Committee Chairman Tim Scott have called for public hearings, and stalled bills like the Remote Access Security Act are gaining renewed interest, The Verge adds. House Speaker Mike Johnson cited a “growing consensus in Congress on limiting chip exports to China,” per Reuters via The Verge.
Nvidia’s longtime chief security officer David Reber and its head of product security stepped down in August 2026, with Reber citing a young-onset Parkinson’s diagnosis after four years as CSO, TechCrunch reports. His replacement, Saša Zdjelar, previously spent 17 years at ExxonMobil and time at Salesforce and ReversingLabs, per TechCrunch. The New York Times cited five Nvidia employees describing Huang’s involvement with a vice president who reported to him, and Reuters later reported a second relationship, TechCrunch notes. Retired intelligence officer Nicholas Eftimiades warned that security teams are often underfunded and CEOs frequently resist spending beyond compliance, while IMSI catchers costing about $1,000 can augment human intelligence, TechCrunch adds. Megaspeed maintains Alice Huang hasn’t worked there since August 2025, and rumors persist about Jensen Huang seeking a CEO return, The Verge says.
Commerce Secretary Howard Lutnick is drafting new rules affecting Southeast Asian data centers, some of which are clients of his former firm Cantor Fitzgerald, creating potential conflicts, according to New York Times. Lutnick has sparred with other Trump officials over restrictions on Nvidia chip flows to data centers serving Chinese customers globally, New York Times reports. The A201 data center in Malaysia, set to open next year with hundreds of millions in Nvidia’s most advanced servers, plans to supply a Chinese tech giant legally by operating outside China, New York Times notes. Its operator is a Cantor Fitzgerald client, and Lutnick has advocated a less stringent control regime that would increase US oversight but preserve more of Nvidia’s international business, per New York Times.
Why it matters: The scandal exposes supply chain vulnerabilities and insider threat gaps that could reshape global AI chip distribution and congressional oversight of semiconductor exports.
Anthropic Claude Fable 5.2 code brushstroke painting
Anthropic’s new language model Claude Fable 5.2 creates complex, photorealistic paintings by using code to place hundreds of thousands of simulated brushstrokes on a virtual canvas, differing from diffusion models like Midjourney by modeling bristle texture, paint behavior, and mistakes it must paint over, according to The Decoder (Simon Willison). Developer Jake Eaton, who works at Anthropic on language model values and safety, demonstrated on X that Fable 5.2 generated a shopping cart in a parking garage using roughly 700,000 brushstrokes without any external image model or pre-rendered sources, The Decoder (Simon Willison) reports. Each painting is a Python program of 1,000 to 2,000 lines driving a simulated paint engine, resembling agentic image editing where the model picks tools and evaluates results, per The Decoder (Simon Willison).
After criticism that the sibling model Opus 5.5 merely imitated existing art, Eaton had Fable 5.2 develop its own artist personas with fictional backstories and consistent styles, The Decoder (Simon Willison) notes. Anthropic engineer Michael Graczyk shared a chat log and video of Fable 5.2 painting a bird at sunrise using exactly 111 strokes over 21 seconds, with the model outputting each stroke’s color, width, and waypoints — including a deliberate seven-pixel wobble to mimic human hand tremor, The Decoder (Simon Willison) details. The model self-corrected between passes: when the bird’s back appeared too thin, it laid a bolder stroke that turned stray bristle marks into ruffled feathers, and in a 280-stroke dusk scene it repainted accidentally smeared objects and positioned figures so their inner shoulders touched as prompted, per The Decoder (Simon Willison).
Fable 5.2’s step-by-step planning and physical self-correction extend beyond virtual canvases to long-horizon technical tasks, as does Mythos 5.2 for verified security and biology researchers, according to The Decoder (Simon Willison). Internal tests show Fable 5.2 solved two open problems in black hole physics, designed three working electronic devices with circuit board manufacturer Diode, and used more than 1,000 sub-agents to program its own laptop operating system in eight days, The Decoder (Simon Willison) reports. AI researcher Aris Zhu tested Fable 5.2 against OpenAI’s GPT-6 Astra on physical robot arms painting Van Gogh’s “The Starry Night”; Anthropic reports a 64.9 percent score on its internal RoboDojo Sim benchmark versus 29.8 percent for GPT-6 Astra, though real-world fine motor control remains limited and Astra painted faster, with only OpenAI’s model managing to stand a domino on end by slowly loosening its gripper, per The Decoder (Simon Willison).
Why it matters: Claude Fable 5.2’s code-driven, physics-aware approach enables self-correcting, long-horizon reasoning that bridges creative generation and robotic control, signaling a shift toward embodied AI agents.
Tools mentioned in this briefing
- Generative AI Book on Amazon — The story describes an AI model that creates images by writing code to simulate brushstrokes, a technique explored in generative AI books. See current options and reviews on Amazon. View on Amazon
As an Amazon Associate I earn from qualifying purchases.
The Nvidia scandal underscores how personal ties can undermine export controls, while Anthropic’s brushstroke model shows code-driven simulation advancing both artistic fidelity and robotic dexterity.
Also: Türkçe podcast.
This briefing is produced with AI assistance from the linked primary sources. Narration uses a synthetic AI news-anchor voice, built with Higgs Materials licensed from Boson AI USA, Inc.

Ali YAZICI is a Senior IT Infrastructure Manager with 15+ years of enterprise experience. While a recognized expert in datacenter architecture, multi-cloud environments, storage, and advanced data protection and Commvault automation , his current focus is on next-generation datacenter technologies, including NVIDIA GPU architecture, high-performance server virtualization, and implementing AI-driven tools. He shares his practical, hands-on experience and combination of his personal field notes and “Expert-Driven AI.” he use AI tools as an assistant to structure drafts, which he then heavily edit, fact-check, and infuse with my own practical experience, original screenshots , and “in-the-trenches” insights that only a human expert can provide.
If you found this content valuable, [support this ad-free work with a coffee]. Connect with him on [LinkedIn].





