D.A.D. joke of the day: My kid asked if I've been using AI to answer his texts. I replied, "Great question — and you're absolutely right to ask!"
What's New
AI developments from the last 24 hours
Under Fire Over Nvidia, Commerce Issues a Tougher Chip Rule and Defends Its Megaspeed Probe
Regulation & Policy China & Geopolitics Nvidia
The Commerce Department spent Monday on defense. It issued a rule to stop advanced American chips from reaching China through third countries: licenses are now required to ship them to Singapore, Malaysia, Indonesia, Thailand and Brazil unless the buyer is American, and for data centers anywhere that rent large amounts of computing power to Chinese companies. The same day, Commerce and the Justice Department took the unusual step of publicly defending their handling of Megaspeed, the Singapore company at the center of the Nvidia scandal, in a statement the White House shared.
The statement says Commerce opened its investigation in 2024, interviewed Megaspeed's CEO, Alice Huang, in June 2025, and handed the criminal case to Justice; that same month, Nvidia stopped selling to Megaspeed. It did not address The New York Times' report that the case moved after Nvidia's top lawyer contacted investigators (D.A.D., November 10).
The rule still leaves gaps: it covers only five countries, and Chinese companies can still buy advanced chips inside the United States or rent them at American data centers. Chris McGuire, who oversaw chip export controls at the National Security Council and is now at the Council on Foreign Relations, called the rule a real step forward but found the defense hard to square. Megaspeed kept buying chips after Nvidia cut it off, he noted, citing a report that it imported about $787 million worth in November 2025. Commerce never put it on the Entity List, the government's tool for blocking exactly this kind of buyer, even though State, the Pentagon, Treasury and the U.S. trade representative favored doing so for months, according to Treasury Secretary Scott Bessent. "But Commerce has the pen," he said. And the rule itself came with an extraordinary admission: Commerce said Treasury and the National Security Council wrote it. "[This] is remarkable and unprecedented. I have never seen a BIS rule that was not written by BIS. That implies that Commerce was dragged kicking and screaming over the finish line by the White House and Treasury."
Why it matters: Washington is finally moving to close the side doors that let Chinese AI labs use American chips. But Treasury wrote Commerce's rule, and Commerce is publicly defending a record its critics say let Megaspeed keep buying. The department responsible for keeping America's best chips out of China is under pressure from its own administration, from Congress and from the scandal that toppled Nvidia's founder.
Sources: Bureau of Industry and Security · Commerce and Justice Departments' joint statement · Chris McGuire on X · The New York Times
Carnegie Mellon Puts Full Computer Science Courses Free on YouTube
Carnegie Mellon is releasing full semesters of its computer science lectures free on YouTube through a new platform called CS Learn. Courses cover AI, robotics, systems and other fields. One course premieres daily during launch week, with more arriving in January. CMU faculty member Phillip Compeau said student-built tools let a small team automate production while still reviewing lectures. The university plans to add exercises, AI tutoring, and paths to official credentials via CMU Online. Compeau acknowledged MIT got there 25 years ago. One Hacker News commenter recalled that accessibility liability once pushed many universities to pull similar YouTube courses.
Why it matters: As AI reshapes which skills employers value, elite computer-science training is becoming freely accessible. That lowers the barrier for anyone retooling for an AI-driven job market, not just enrolled students.
Discuss on Hacker News · Source: cs.cmu.edu
Windows 11 Logs Which AI Coding Tools You Use, Researcher Finds
A researcher used AI to comb through Windows binaries and found that a system component called MissionControlAggregator.dll records which coding tools are used on a PC. These include Claude, Copilot, Cursor, Codex, Python, and other developer software. The data is saved locally and, unless a user has opted out of Microsoft's optional diagnostic data (on by default), uploaded through Windows's telemetry system. The AI-agent checks sit behind an internal feature flag called CLICodingAgentUsage.
Why it matters: If your teams run Claude, Cursor or Codex on Windows, Microsoft may see that usage by default. IT leaders may want to review diagnostic-data settings. It also gives a platform owner visibility into rivals' adoption.
Discuss on Hacker News · Source: xusheng.dev
What's Controversial
Stories sparking genuine backlash, policy fights, or heated disagreement in the AI community
Inside Nvidia, a Long-Shot Push to Bring Back Jensen Huang
Some of Nvidia's own people want Jensen Huang back. A senior employee, developer technology director Nader Khalil, wrote to the board and top executives warning that the company could lose "irreplaceable talent" if Huang isn't reinstated as CEO, The Wall Street Journal reported. The letter, also sent to Huang's roughly 60 direct reports, argues he deserves the benefit of the doubt while the board investigates his relationship with Alice Huang. Outside the company, a "Stand With Jensen" petition whose signers include Elon Musk and Epic Games CEO Tim Sweeney has drawn more than 180,000 signatures, Reuters reported; its organizers say about 8,900 signers identify as Nvidia employees, a figure that can't be verified. Nvidia hasn't responded publicly.
The odds look long. Huang resigned from the board as well as the CEO job, and Treasury Secretary Scott Bessent has called for an outsider to replace him (D.A.D., November 16).
Why it matters: The push shows how much of Nvidia's identity was built around one person, and how unsettled the company remains as interim CEO Colette Kress steers it through Tuesday's earnings report, a CEO search and a pullback from its role bankrolling the AI boom.
Sources: The Wall Street Journal · Reuters, via CBC · Spectrum News
What's in Academe
New papers on AI and its effects from researchers
Synthetic Survey Respondents Flatten Real Differences Between Groups
AI models increasingly answer surveys on behalf of real panelists. A new study tested whether feeding them each panelist's demographic record, such as age, education and ethnicity, makes those answers more faithful. Across two U.S. surveys and 14 models, researchers found the personal record added little to individual predictions. For population estimates in open-weight models, census data or a synthetic persona kept most of its benefit. Worse, when comparing groups, models leaned on demographic labels rather than real data, often flattening or reversing actual attitude differences, particularly around religion and ethnicity.
Why it matters: Companies and researchers increasingly use AI to simulate survey respondents instead of polling real people, but this suggests those simulations can mask or distort genuine group differences while creating a false sense of personalization.
Reddit Health Forums Show Rising Privacy Fears About Medical AI
Privacy & Surveillance Health & Medicine
A study analyzing 6,514 Reddit posts from 14 medical communities found users increasingly discuss security and privacy worries about healthcare AI. Posters flagged how EHR integrations, cloud platforms, and automated clinical tools expand the ways their data can be exposed. Researchers identified four recurring attitudes. They range from questioning accountability for AI harms and cautious acceptance with safeguards to skepticism and resignation to growing data exposure. Discussion volume rose alongside major AI deployments, privacy incidents, and new regulations.
Why it matters: As hospitals and health systems adopt AI tools, patient trust—not just clinical accuracy—will shape whether those systems actually get used.
Legal Scholars Propose Rules for Who's Liable When AI Agents Overstep
A legal paper tackles a question companies deploying AI agents will increasingly face: who's on the hook when an AI system acts outside its intended authority? The proposed framework attributes the overstep to whoever delegated power to the agent, measured against what that party documented (the agent's intended "perimeter") and how much control it actually exercised. The paper suggests companies keep a formal "delegation record" proving those boundaries, and analyzes the rule under EU and Italian law using the July 2026 OpenAI and Hugging Face incident as a case.
Why it matters: As businesses hand AI agents more autonomy to book travel, send emails or make purchases on their behalf, clear documentation of what an agent is authorized to do may become the difference between a contained mistake and a costly legal liability.
Model Upgrades Make AI Peer Reviewers Sharply Harsher
Researchers tested twenty leading models from OpenAI, Google and Anthropic as peer reviewers on ICLR 2027 submissions while they were still under review. Newer versions of the same model judged identical papers far more harshly than older versions did. The share of papers placed on the accept side fell by 43 to 79 percentage points between versions. No paper ever moved up. Verdicts shifted even when a critique's technical points stayed the same. Model scores tracked human reviewers only weakly, though no worse than a third human reviewer did. Claude Opus 5.5 also named the true first author of 27 already-decided papers from titles alone, suggesting memorization can skew tests on older papers.
Why it matters: As journals, conferences, and companies increasingly lean on AI to screen submissions, grants, or job applications, this suggests the verdict may depend more on which model version you happened to use than on the quality of the work itself.
Shoppers Follow AI Picks, but Store Analytics Mostly Miss It
A new study tracked what happens after people ask AI assistants for shopping recommendations, matching those conversations to actual browsing at twenty major retailers. Shoppers followed through. Among those new to a store, visits to the named product within a day rose roughly 100-fold. On days with a product-page visit, 44% showed a product from the shopper's own AI answer. But 89% of these visits carried no sign of the assistant. Most arrived via web search, so stores logged them as ordinary traffic.
Why it matters: Retailers' marketing dashboards are likely undercounting how much AI assistants are already steering shoppers, which could lead companies to misjudge where to invest in advertising, SEO, and product-placement strategy.
What's On The Pod
Some new podcast episodes
How I AI — How I fight token fraud as a solo founder
AI in Business — How IT Leaders Get More Operational Value From Their ServiceNow Investment - with Assaf Resnick of BigPanda