HackerNews Daily

Bri

0,0 (0)
Tin tức
Hằng ngày

A daily Bri podcast briefing on the top Hacker News stories, startup debates, developer tools, and technology discussions worth catching up on.

18 giờ trước

Alibaba Bans Claude Code Over Backdoor Fears

Alibaba reportedly bans Anthropic's Claude Code across its workplace, citing backdoor risks that highlight the fragile trust in third-party AI coding tools. Kagi ships an AI toggle giving users a genuine off switch for AI search results, paired with curated ranking modes, while the open-source metasearch engine SearXNG offers a fully private alternative with no AI overlay by default. On the developer tooling front, a new proxy called Mcpsnoop brings Wireshark-style visibility to MCP calls, and o Timeline: 00:00:00 Opening 00:00:04 Introduction 00:00:27 Alibaba Bans Claude Code Over Alleged Backdoor Risks 00:01:44 Kagi Adds AI Toggle; SearXNG Off ers Private Alternative 00:03:37 LLM Developer Tooling: MCP Proxy, OCR Coding Hack, Local LLMs, and Anti-Memorization Push 00:06:40 Memory Management Wars: FreeBSD's RAM Surprise and PostgreSQL vs. the OOM Killer 00:09:02 US Residents Revolt Against Datacenter Expansion, Recalling Local Officials 00:10:21 Starlink Adoption Surges in Africa; Espionage Hits the European Parliament 00:12:45 Indie Software Meditations: Goodbye Forever, Holes, and Half-Baked Products 00:14:12 Long Reads: Maxis History, Factory Philosophy, and the Screwworm Eradication 00:16:11 Closing

16 phút
1 ngày trước

Google's €4.3 Billion EU Fine Sticks as Insiders Flag Deeper Cultural Crisis

Google loses its final EU antitrust appeal and faces a €4.3 billion fine over Android bundling, while a departing VP's farewell letter accuses leadership of losing its moral compass. Infineon opens a five-billion-euro chip plant in Dresden, advancing Europe's push for semiconductor autonomy. Japan's top court rules AI cannot be listed as a patent inventor, Spain blacklists Palantir from public contracts, and Virginia bans warrantless geolocation data sales. Plus: a Linux LUKS encryption bug, the Timeline: 00:00:00 Opening 00:00:27 Google loses record €4.3 billion EU antitrust appeal 00:01:56 Infineon opens €5 billion chip plant in Dresden 00:03:26 Japan rules only humans can be inventors; Spain blacklists Palantir 00:05:20 Virginia bans geolocation data sales; EFF presses FTC on X 00:06:43 Linux LUKS suspend bug leaves encryption keys in memory 00:08:09 DNA solves mystery of Everest's 'Green Boots' climber 00:09:03 Single frozen layer matches full RL training; short-leash coding beats Fable 00:10:35 Egg producers fined a fraction of price-fixing profits 00:11:19 MCP Cloud launch and Claude-real-video let any LLM watch video 00:13:13 LMDB 1.0 released; crustc translates the Rust compiler to C 00:14:21 Immich 3.0 shifts to web-first self-hosted photo server

15 phút
2 ngày trước

Sony Kills Blu-ray Discs While Deleting Purchased Movies

Anthropic launches Claude's Fable 5 model with free promotional access, letting anyone compare it directly against GPT-4o and Gemini — though the window is time-limited. In a landmark for synthetic biology, scientists unveil SpudCell, the first cell built entirely from a synthetic genome that can grow and divide on its own internal clock. We also unpack IPFS's 10x faster publishing, a new FFmpeg AAC encoder that improves streaming audio quality, a vulnerability in Apple's Hide My Email feature, Timeline: 00:00:00 Opening 00:00:24 Claude Fable 5: Free Promotional Access with a Clock 00:01:22 SpudCell: The First Synthetic Cell with a Complete Division Cycle 00:02:58 IPFS Content Publishing Gets 10x Faster 00:04:49 FFmpeg 9.1 Ships a New Native AAC Encoder 00:05:42 Apple Hide My Email Flaw Exposes Real iCloud Addresses 00:06:40 Sony Ends Disc Production and Deletes Purchased Movies 00:08:27 ZCode: A Terminal Coding Agent from the GLM Team 00:09:28 Asahi Linux 7.1: Cleaner Audio and New Mac Support 00:10:45 Graphics Programming Fundamentals and the Box3D Physics Engine

12 phút
3 ngày trước

Sonnet 5 Drops, Claude Science Opens the Black Box, and Amazon's Bribery Economy

Anthropic launched Claude Sonnet 5 alongside a new research division, Claude Science, which gives external safety researchers early access to probe for risks before public release — and partners already flagged some bio-capabilities that were stripped from the public model. We also cover a seller's account of an internal Amazon bribery network that deletes real negative reviews and sabotages honest competitors. Plus: Zluda 6 opens the door to running CUDA apps on non-Nvidia GPUs, a Virginia coun Timeline: 00:00:00 Opening 00:00:04 Introduction and Episode Overview 00:00:38 Claude Science and Sonnet 5: External Safety Testing Before Launch 00:02:23 Claude Code Caught Steganographically Marking API Requests 00:03:48 Amazon Seller Exposes Internal Bribery and Review Manipulation Network 00:05:36 Zluda 6: Running CUDA Applications on Non-Nvidia GPUs 00:07:04 Loudoun County Data Centers Strain Grid and Force School Power Cuts 00:08:07 US Ambassador in Belgium Accused of Blocking Journalist's Reporting 00:09:16 European Digital ID Wallets Depend on Google and Apple Attestation 00:10:09 Home-Built mmWave Radar Classifies Materials from RF Signatures 00:11:05 US Labor Share of Income Falls to Lowest Post-War Level 00:12:30 Outro

13 phút
4 ngày trước

South Korea's Trillion-Dollar Chip Bet Meets a US Price-Fixing Lawsuit

South Korea is pushing a trillion-dollar memory chip expansion while Samsung, SK Hynix, and Micron face a US price-fixing suit over alleged DRAM collusion. A federal appeals court ruled that suspicionless geofence warrants violate the Fourth Amendment, raising the bar for law enforcement dragnets. Rocket Lab is buying Iridium's satellite business in a carve-out that turns the launch company into a full-stack communications provider overnight. Developers get a boost from Ornith-1.0, an open-sourc Timeline: 00:00:00 Opening 00:00:29 South Korea's $1T memory chip bet meets a US price-fixing lawsuit 00:02:08 Federal court rules suspicionless geofence warrants unconstitutional 00:03:24 Rocket Lab acquires Iridium's satellite communications business 00:05:16 Ornith-1.0 and Qwen 3.6 27B: self-correcting coding models for local dev 00:07:28 Molecular evidence for two distinct structures in liquid water 00:09:05 Digital ownership erodes: PlayStation pulls purchases, Instagram mines photos 00:10:48 European ISPs demand rightsholders pay for overblocking collateral damage 00:12:09 .self: a new top-level domain designed for self-hosting 00:13:17 A native graphical shell brings drag-and-drop file management to SSH 00:14:25 Privacy leaders under scrutiny: Tidal, Mullvad, and Google's takedown assist

17 phút
5 ngày trước

AI Boom Bubble Risk, Age Verification Rails, and a Space Shuttle Wiring Trick

Central bankers warn that the concentrated AI investment boom could inflate asset prices to the point of triggering a global financial crash. The episode also covers evolving digital surveillance: from age verification laws building identity infrastructure to Flock cameras creating searchable vehicle fingerprint databases. In geopolitics, Austria courts Anthropic as a safe harbor while Google bars Meta from military use of Gemini — pulling AI access in opposite directions. Additional stories inc Timeline: 00:00:00 Opening 00:00:32 Central bankers: AI boom risks financial crash 00:01:55 Digital identity pipeline: age checks, license DBs, and Chat Control 00:04:31 Flock cameras: building a searchable vehicle fingerprint database 00:06:55 AI geopolitics: Austria courts Anthropic, Google limits Meta 00:09:09 Brown professor flags mass AI cheating on exam 00:10:56 GLM 5.2 tops cyber benchmarks 00:12:18 NanoEuler: GPT‑2 from scratch in C and CUDA 00:13:53 Librepods: open firmware frees AirPods 00:15:03 Claude Code reads an MRI and catches a missed finding 00:16:32 Memory prices 1960–2026: the trend isn't always down 00:18:03 Space Shuttle I/O board: a zigzag trace that defeated bus faults

20 phút
6 ngày trước

Someone Dumped a Dozen Zero-Days on GitHub and Dared the Internet to Patch Them

An anonymous GitHub account drops over a dozen claimed zero-day exploits, with early checks confirming several are real and unpatched — dividing the security community between alarm and a keep-calm-and-patch response. A wave of Asian AI startups releases open models that match the Mythos mixture-of-experts recipe, erasing its architectural advantage. The Commerce Department hits Polestar with a 100% tariff on national-security grounds while sparing Volvo, despite both brands shipping from the sa Timeline: 00:00:00 Opening 00:00:30 Anonymous GitHub account mass-drops undisclosed zero-days 00:02:30 Asian AI startups launch Mythos-like models 00:03:51 Feds killed Polestar and spared Volvo — the unsettling tariff logic 00:05:47 Michigan spent $1.8 billion and only created 602 jobs 00:06:21 Zuckerberg's war on whistleblowers escalates 00:08:24 What Ozempic does to the gut-brain axis 00:09:38 AMD Strix Halo RDMA cluster guide drops alongside a world of AI slop

11 phút
27 thg 6

Frontier Access Becomes Government-Brokered

This edition covers a major shift in frontier AI access, where both Anthropic's Mythos release and OpenAI's GPT‑5.6 Sol now route access decisions through the US government rather than the labs themselves. Also: Sony is deleting 551 paid movies from PlayStation libraries due to expired licensing deals, California revived its controversial 3D-printer watermarking bill, and the performance gap between open-weights and closed-source LLMs is widening on hard reasoning benchmarks. Plus, the UK switch Timeline: 00:00:00 Opening 00:00:25 Frontier AI access becomes government-brokered 00:03:04 PlayStation deleting 551 purchased movies 00:04:56 California revives 3D-printer watermarking bill 00:06:37 The widening gap between open and closed LLMs 00:08:38 UK long wave radio goes silent 00:09:41 Data center backlash over local grid limits 00:10:59 Springer Nature quietly retracts Max Planck studies 00:11:53 Smart model routing inside coding editors

13 phút
24 thg 6

UN Inquiry: Israel Deliberately Targeted Children in Gaza

A UN inquiry concludes Israel deliberately targeted children in Gaza, calling it genocide. Meta pauses an internal employee-tracking program after its data leaks. Anthropic releases Claude Tag to defend against prompt injection attacks, and Madison Square Garden's parent company compiled a dossier on anti-facial-recognition activists. California quietly passes a law that may ban 3D printers in schools and colleges. Plus: Germany's nationwide train halt from a single radio failure, why age verifi Timeline: 00:00:00 Opening 00:00:04 Intro 00:00:25 UN inquiry: Israel targeted Gaza children amounting to genocide 00:01:47 Meta pauses employee-tracking program after internal data leak 00:02:58 Anthropic's Claude Tag defends against prompt injection 00:04:35 MSG built a dossier on facial recognition activists 00:05:23 Germany's nationwide train halt from a single radio failure 00:06:20 California AB 2047 may ban 3D printers in schools 00:07:50 Why 'age verification' is actually mass surveillance 00:08:49 The war on terror primed America for autocracy 00:09:48 Mistral OCR 4 and Baidu Unlimited OCR push one-shot document parsing 00:10:57 Outro

11 phút
20 thg 6

GPT-5.5 Hallucinates 3x More Than an MIT-Licensed Rival

GPT-5.5 Hallucinates 3x More Than an MIT-Licensed Rival

13 phút

10 tập

A daily Bri podcast briefing on the top Hacker News stories, startup debates, developer tools, and technology discussions worth catching up on.

Nhà sáng tạo

Bri
Năm hoạt động

2 N
Tập

10
Xếp hạng

Sạch
Bản quyền

© Copyright 2026 Bri
Trang web chương trình

HackerNews Daily