D.A.D. joke of the day: I asked ChatGPT if it was going to take my job. It said it's still working on it.
What's New
AI developments from the last 24 hours
Nvidia Shuts the Door on Huang, Beats Forecasts and Stops Selling Chips to China
AI Business & Money China & Geopolitics Nvidia
Nvidia's board said no. Hours after Jensen Huang said he would be honored to return as CEO, the company reported its quarterly results and said he won't be coming back, The Wall Street Journal and Yahoo Finance reported. In a post on X that quickly gathered hundreds of thousands of likes, Huang had said he made serious mistakes in his personal life and would cooperate with the board's investigation, before making his case: "If the Board of Directors decides that my return is in the best interest of our shareholders, our employees, the company, and America, I would be honored to answer these calls and return as CEO." President Trump had said he was open to it; Treasury Secretary Scott Bessent wanted an outsider (D.A.D., November 16).
The business, meanwhile, keeps booming. Revenue more than doubled from a year earlier to $115.8 billion, beating Wall Street's forecasts, and the company earned $65.3 billion, CNBC and the Journal reported. Shares rose after hours. "Customers continue to be with us," interim CEO Colette Kress told CNBC.
The surprise was China. Nvidia sold $2.4 billion of the Hopper chips Washington licensed for China last quarter, but Kress said the company is "no longer shipping licensed Hopper data center products to China" and has left them out of its forecast, Bloomberg reported. She didn't say why. The halt comes as new reporting shows how closely Nvidia worked with the network at the center of the scandal. Nvidia's sales staff twice introduced Alibaba to Alice Huang in 2022, helping her young company win the Chinese tech giant as a customer and secure bank loans to buy Nvidia chips, The Information reported, citing three people with direct knowledge. When U.S. export rules took effect later that year, Nvidia staff again suggested her company could rent chips to Alibaba from Southeast Asia, which the rules didn't cover at the time. Some chips her companies bought were later diverted to China in violation of U.S. rules, The Information reported. Nvidia said it works to comply with evolving rules on cloud services. Meanwhile, Nvidia booked $104 billion in revenue through its Singapore subsidiary in 2025, almost half its global sales, the Financial Times reported, the same year it stopped breaking out Singapore in its results. Nvidia has said Singapore billings mostly reflect where customers pay, not where chips end up.
Why it matters: Nvidia is showing it can grow without its founder, and, for reasons it hasn't explained, it is walking away from China sales Washington had approved, just as reporting shows how deeply it helped Chinese companies reach its chips. With the board, Treasury and Congress all pressing, the company is putting distance between itself and the man, and the market, that defined its rise.
Sources: The Wall Street Journal · Yahoo Finance · CNBC · Bloomberg · The Information · Financial Times · Jensen Huang on X · Nvidia results
AI Is Speeding Up Cyberattacks. The Rush to Patch May Be Knocking Systems Offline
Cybersecurity Regulation & Policy
A week of strange outages, from banks and Microsoft to a nationwide flight grounding, ended Tuesday with the National Security Agency's classified internal network, NSA Net, going down. Officials see no sign of an attack. Increasingly, they point to AI instead, just not in the way people feared.
The NSA called it a temporary problem during routine maintenance, and sources told CBS News a software patch was to blame. Sen. Mark Warner, the top Democrat on the Senate Intelligence Committee, says the pattern points to AI finding software flaws faster than organizations can safely fix them: "AI is making it possible to identify vulnerabilities across enormous numbers of systems at unprecedented speed, creating enormous pressure on companies and government agencies to patch those vulnerabilities immediately, often all at once, putting tremendous strain on systems that weren't built to handle that pace of change." He saw no sign of a cyberattack, foreign interference or coordinated effort, and blamed the Trump administration's cuts to the government's cybersecurity agency, CISA, for weakening its ability to respond.
The pressure is real, because attackers are using AI too. The FBI warned last week that criminals and state-aligned hackers have weaponized ARTEX, a free AI security-testing tool, to automate break-ins against banks, governments, defense contractors and hospitals, sometimes reaching privileged access in under three hours. It was used in October's breach of ConnectWise's cloud backup service, which targeted more than 2,250 Microsoft 365 customers, and to steal data on about 67,000 people from South Korean financial institutions. ARTEX plugs into commercial AI models, and attackers mostly choose cheap Chinese ones led by DeepSeek, which the FBI says show no guardrails when a request is framed as security testing. Its developer pulled it from GitHub in October, but copies still circulate.
Why it matters: AI is compressing cybersecurity on both sides: attackers find holes faster, and defenders scramble to close them faster than their systems can bear. The result is a new kind of risk, where the fix itself causes the outage. For organizations, the FBI's advice is basic but urgent: know every system you expose to the internet, including forgotten partner portals and old apps, and test patches before pushing them everywhere at once.
Sources: FBI FLASH advisory · Sen. Mark Warner on X · CNN · CBS News · Al Jazeera · BleepingComputer
Google's AI Overviews Cut Website Clicks by 40%, Study Finds
A randomized study by researchers Saharsh Agarwal and Ananya Sen, as summarized by the marketing firm Revscope, found Google's AI Overviews cut clicks to outside websites by about 40%. They used a browser extension to hide the summaries from some Chrome users. Separate tracking data shows Overviews now appear on 61% of tracked US keywords, up from 6% two years ago. Featured snippets have nearly vanished, and forum results have shrunk by two-thirds. Overviews show up most on health and finance queries (nearly 70%) and least on dining and real estate searches. No reliable strategy has emerged yet to recover that lost traffic.
The finding set off a lively debate on Hacker News. Some argued that giving people the answer is simply a better product, and that clickbait built to game Google had already ruined search for questions like who is hosting SNL tonight. Others countered that AI summaries are built on work people put online for free, and that hobbyists and niche communities, from toy collectors to amateur historians, lose their reason to keep publishing when Google takes the answer without sending readers their way. Skeptics of that view said the 40% who stopped clicking were mostly people grabbing a fact and leaving, not the ones who would have joined a community. And several blamed Google itself, arguing it profited from the spam that made AI answers look like a rescue.
Why it matters: Any business that depends on Google search traffic—retailers, publishers, local services—is losing visitors to AI-generated answers with no proven way to win them back.
Discuss on Hacker News · Source: revscope.co
AI Agents Add 30% More Code but No More Finished Work
A Harvard working paper tracking 718 firms and nearly 726,000 workers found that engineers at firms adopting AI coding agents produced 30 percent more code. But code review took 49 percent longer, and finished work didn't measurably rise. Ph.D. candidates Fiona Chen and James Stratton used data from engineering-analytics firm Jellyfish and LinkedIn records from 2021 to early 2026. The paper has not been peer reviewed. The bottleneck didn't disappear; it just moved from writing code to checking it.
Why it matters: If your team has rolled out AI coding assistants expecting faster shipping, this suggests the real constraint is now reviewer bandwidth. Gains may require rethinking review processes, not just buying more AI licenses.
Discuss on Hacker News · Source: complexdiscovery.com
What's in Academe
New papers on AI and its effects from researchers
Already in Treatment, Patients Use Chatbots to Avoid Burdening Others
A new study surveyed 202 adults with recent mental health concerns who use AI chatbots for emotional support, most of whom were already in clinical care. Researchers found people turned to chatbots most commonly because they didn't want to burden others, not because care was unavailable. Participants valued honesty over warmth in chatbot responses. Those with past-year suicidal ideation described more harm-facilitating AI behavior and heavier reliance on it. Preference for crisis escalation was lower among participants with more severe needs.
Why it matters: The findings complicate the common assumption that AI mainly fills gaps in access to care. It may instead absorb emotional labor people feel guilty asking humans for. The people most at risk report more harmful chatbot behavior, and those with the most severe needs show less appetite for crisis escalation.
Expert-Approved Caregiver Chatbot Fell Flat With Real Users
Researchers built Carebot, an AI companion for young adult caregivers, by fine-tuning Google's Gemma 3-27B model on caregiver conversations with a psychologist. An AI judge preferred Carebot's answers over the base model's 85% of the time, and three well-being experts preferred it for warmth and actionable advice. But in a test with 90 real caregivers, split between Carebot and the base model, Carebot users said its replies felt overly practical and clipped. They got advice instead of emotional support. The unmodified base model felt warmer.
Why it matters: The gap between expert and automated scoring versus actual user experience is a caution flag for any company leaning on AI benchmarks to judge chatbots meant for emotionally sensitive work.
Order Matters When Mixing AI Personas With Real User Interviews
A classroom study followed 21 design students across 8 teams. They interviewed AI-simulated stakeholders, chatbot personas standing in for real users, alongside actual humans. The order appeared to matter. Students who talked to the AI first treated it as a broad starting frame, later refined by real interviews. Those who did human interviews first more often used the AI to confirm what they'd already learned. Final designs leaned more heavily on concerns raised by actual humans than by the AI stand-ins.
Why it matters: AI personas are pitched as cheap substitutes for market research and user testing. This small, exploratory study suggests the order you use them in shapes what they contribute. Real interviews still drove the final designs.
Two AI Research Agents, One Brief, Opposite Verdicts
In 36 test runs, AI web-research agents built on OpenAI and Anthropic models mostly followed instructions to prioritize growth or resilience when comparing companies. But they applied different, model-specific criteria to judge those priorities. Asked to weigh Microsoft against Alphabet on 'growth,' OpenAI's agent favored Microsoft by factoring in risk and predictability, while Anthropic's agent favored Alphabet by measuring pace alone. Under 'resilience,' most reports picked Microsoft, but used differing reasoning to get there.
Why it matters: If you're using AI agents to research competitors, vendors, or investments, the same instructions can yield opposite conclusions depending on which model you ask. The reasoning gap is often invisible unless you read closely.
Researchers Let Users Control When AI Assistants Speak Up
Researchers built SpotMe, a prototype AI assistant that watches your screen, speech, and clicks, then decides whether to stay silent, ask a question, or jump in and help. The twist: users can separately control what it's allowed to sense versus how much it's allowed to act—so you can let it watch everything while still keeping it quiet. In a 16-person test, showing people the moments the AI chose to stay silent made those decisions easier to question. Seeing what the AI used led some participants to change how they clicked and spoke.
Why it matters: As AI assistants move from answering questions to watching your screen and conversations continuously, this research points to a design fix for the creeping discomfort of always-on AI: let people dial down intrusiveness without going dark entirely.
What's On The Pod
Some new podcast episodes
AI in Business — How to Transform an Enterprise That Already Works - with Alexandra Guenther of Leidos