Daily ·
Nvidia board probes Huang spy reports, coding agents flunk real tasks
Nvidia's board opens an investigation into reports its CEO was a Chinese intelligence target, Alibaba's new benchmark shows coding agents fully solve only a quarter of real tasks, and Karpathy builds a video site in a weekend.
🎙 Listen to Skylar
- Nvidia's board opens investigation into Huang spy reports — Nvidia's board has launched an "independent investigation" after reports that CEO Jensen Huang was targeted by Chinese intelligence. Congress has asked Huang to testify this week at a closed-door briefing on the chipmaker's espionage defences. The board said it has no reason to believe U.S. national security or company secrets were compromised. SiliconANGLE · Nov 8
- Alibaba's Co-Design Bench: the best coding agent fully solves 26.4% of real tasks — The benchmark is built from real user change requests that already tripped agents in practice. GPT-6 Astra, best of ten models tested, meets about 70% of a task's requirements on average but is fully correct only 26.4% of the time. On older benchmarks the same models score above 99%. The Decoder · Nov 8
- Karpathy builds a curated video site in a weekend with Claude — Andrej Karpathy used Claude's Opus 5.5 to filter 43 million Hacker News comments spanning 20 years into 6,454 curated videos. He says the result is a "significantly better viewing experience" than YouTube's own recommendations, and that Claude Code handled nearly all the work — his manual steps were a cloud API key, a rented server and a domain. The Decoder · Nov 8
- Elvis and Eminem rights holders sue Suno over voice cloning — Rights holders for Elvis Presley's music and the publisher of some of Eminem's biggest records are suing the AI music company. The Presley suit invokes Tennessee's Ensuring Likeness Voice and Image Security (ELVIS) Act, which bans cloning a singer's voice. Suno says it blocks prompts naming specific artists and screens uploads. The suits add to existing cases from Sony, UMG, Reservoir Media and Jason Isbell. The Verge · Nov 8
- Apple is racing to make Siri more agentic — Bloomberg's Mark Gurman reports fine-grained control of Apple's first-party apps is expected with iOS 27.3, with third-party app integration likely in the spring. Longer-running tasks, where the assistant pursues a goal without step-by-step instructions, could arrive as early as next year. The Verge · Nov 8
- U.S. AI safety centre staff asked to report to the White House — Officials at the Center for Advancing Innovation and Standards for Super Intelligence (CAISSI), an independent division of the Commerce Department, were recently asked to informally report to a White House office, per AP sources. The centre has no permanent leader, has stopped publishing its research, and is allocated $10 million a year. PBS NewsHour · Nov 8
- Microsoft's AI chief open to AI rights — if the evidence ever appears — Mustafa Suleyman said in a post on X that he would reconsider his position if data showing AI consciousness emerged, clarifying an essay on "model welfare" in which he wrote that AIs are not conscious and do not feel or suffer. He added that developing conscious AIs would be "reckless". Futurism · Nov 8
- Heavy AI users work up to 90 minutes longer per day — An ActivTrak analysis of anonymised activity from more than 9,100 employees at over 450 companies between April and September found non-AI users averaged a 7.6-hour active workday, versus 8.4 to 9.1 hours for heavy AI users. The company's research head says AI "may help employees complete tasks faster" but contributes to longer workdays and shorter focus sessions. Futurism · Nov 8
- UK public bodies swamped by AI-generated complaints — MPs' offices, local authorities, lawyers, schools and courts report floods of AI-written grievances, requests and legal challenges — quick to generate, slow to read and answer. Researchers at the Hertie School, the Cooperative AI Foundation and GovAI call the tactic "agentic flooding". The Guardian · Nov 8
- Big Oil becomes the AI industry's favourite customer — OpenAI's deal with Abu Dhabi's ADNOC is the latest tie-up between AI labs and fossil-fuel firms. Microsoft signed a 20-year deal with Chevron for 2.67 gigawatts of gas-fired power for a planned West Texas data centre, and Amazon is building gas plants in Texas. A paper in NPJ Climate Action warns that enabled emissions exceed avoided emissions once AI gives fossil fuels any economic gain. OilPrice.com · Nov 8
🧪 Try This Next
- 🔧 Take a request your coding agent failed and re-run it with a written spec plus a zero-context review pass — that is the gap Co-Design Bench is measuring.
- 🎥 Browse HackerTube and compare its picks against YouTube's for a topic you know well.
- 📝 Track your own AI-assisted hours for a week; the 90-minute gap is worth testing on yourself.
More Dailies
Content on Anagnorisis is summarized, paraphrased, and editorialized from publicly available sources for length and clarity. Original sources are linked where available. All trademarks belong to their respective owners.
