aigest

Топ AI-постів сьогодні // 2026.11.08

Тільки свіжі

100

Claude Code sends your email address to websites ↗

Nov 05 · 22:17

100

Anthropic is getting sued over its $200/month Claude plan not delivering the advertised 20x usage ↗

Nov 07 · 20:51

99

Postman silently auto-installs AI plugins and attempts to share all API data ↗

Nov 03 · 17:22

99

A year after the 100% AI mandate nobody understands our code ↗

Nov 07 · 09:46

99

AI is forcing us to pay off 40 years of technical debt ↗

Nov 04 · 11:52

99

Gemini just gave me the list of flags for "dangerous" prompts ↗

Nov 06 · 21:45

98

Stringman, an $800 open-source room scale gripper robot ↗

Nov 03 · 16:47

98

Xiaomi's humanoid robots spotted doing their first 8-hour shift sorting EV parts at a rate of ~ 1 per minute ↗

Nov 03 · 21:16

98

Tell HN: Gemini trains on chat history, even for subscribers (personal accounts) ↗

Nov 04 · 10:10

98

New image model based on Minimax H3, 7× smaller, 5× faster, 8 steps ↗

Nov 03 · 17:09

97

Qwen 3.8 27B on my laptop can interpret model weights, too ↗

Nov 03 · 17:40

97

This guy figured out how to leak your exact location from any video with sound using the Doppler curves of nearby airplanes ↗

Nov 03 · 15:12

97

Designing responsible AI for learning ↗

Nov 03 · 15:25

97

Grok is dead ↗

Nov 03 · 13:02

96

Kimi K3.1 Pro model weights are out ↗

Nov 04 · 15:30

96

Official Statement: Qwen Models Will Stick to Open Source to the End! ↗

Nov 02 · 16:42

96

Nvidia DGX Spark encrypted firmware slows inference speed to 35% on Linux ↗

Nov 04 · 08:46

95

Grok Bot users ran out of a week’s usage overnight, and Elon Musk says the agent was costing too much ↗

Nov 01 · 21:44

95

Ask HN: What's your current AI agentic setup? ↗

Nov 04 · 14:37

95

opus 5.5 XHIGH just deleted all my apps data, passwords and macOS configs ↗

Nov 07 · 20:11

94

Fujifilm to boost tape production 5x as AI fuels archiving demand ↗

Nov 02 · 14:02

94

Viggle Remake is a beast ↗

Nov 07 · 17:21

94

OpenAI seems to be silently re-enabling model training on Codex chats ↗

Nov 03 · 14:53

94

I built a sailing sim with Claude Code and made the frontier models race each other in it ↗

Nov 06 · 17:22

93

Claude Code's "Don't Ask Again" exposes passwords in 1,122 public GitHub files ↗

Nov 07 · 20:45

93

And there go all the spend tracking apps ↗

Nov 05 · 22:33

93

Cuban: Facebook knowingly runs AI deepfake scam ads; docs say $16B fraud revenue ↗

Nov 05 · 23:09

93

ChatGPT leaked an internal prompt ↗

Nov 04 · 22:19

92

Microsoft Researcher John Langford Exits. AI Frontiers Group Declared Finished ↗

Nov 07 · 18:02

92

I built a free, open-source film studio on top of ComfyUI + MiniMax H3: idea to finished short film, on your own PC ↗

Nov 03 · 10:10

92

US tech giants' hidden AI debt swells 90% to $3.2T in 6 months ↗

Nov 03 · 21:26

91

Which LLM is Winning Your Daily use Case Lately? ↗

Nov 03 · 02:22

91

The Little Book of Generative AI Foundations: An Intuitive Mathematical Primer ↗

Nov 03 · 21:40

91

MIT researchers say GPT-6 Astra improved a small Qwen model’s math score from 7.9% to 48.1% by directly editing just 0.016% of its weights, without gradient-based training. ↗

Nov 02 · 20:18

90

AI is dramatically accelerating AI research ↗

Nov 06 · 20:27

90

Almost half of all new Steam games use AI now ↗

Nov 02 · 15:50

90

Anthropic's internal AI use, at retail prices, now exceeds its employee pay ↗

Nov 02 · 16:40

90

Meet Creator Mode in DeepSeek Harness. ↗

Nov 02 · 15:31

89

Anthropic IPO Buyers See Valuation as Low as $1.1T ↗

Nov 06 · 16:54

89

Professor returns from lunch to find AI deleted year’s worth of lecture notes ↗

Nov 07 · 15:57

89

Scaling Git at OpenAI ↗

Nov 02 · 23:47

89

A judge says Claude Max’s “20x more usage” promise could mislead paying subscribers ↗

Nov 07 · 08:02

88

Show HN: A decent text-to-image model trained for ~$20K (Apache 2.0 Weights) ↗

Nov 04 · 16:45

88

Stanford researchers develop a new AI memory system that recalls information from entire conversation histories using just 16 memory tokens. It reduces document-token usage by up to 4,212×, runs 27× faster in a 512-document test, and outperforms existing approaches on 5 of 8 benchmarks. ↗

Nov 06 · 11:17

88

Нове Llama-modchip: 284B DeepSeek on 64GB RAM, no GPU ↗

Nov 08 · 10:22

87

Insane that Deepseek alone accounted for 53% of all tokens on openrouter yesterday, the next 10 labs total under 47%. ↗

Nov 04 · 20:36

87

Progress on provable inference for large language models ↗

Nov 04 · 17:17

87

British fintech Revolut slashes AI coding costs 10x by cutting out Claude and going self-hosted ↗

Nov 03 · 21:05

87

Anthropic releases amended IPO prospectus with third-quarter financials ↗

Nov 02 · 14:38

86

Microsoft files patent for AI that can alter what people say during live broadcasts ↗

Nov 06 · 04:25

86

AI delivers 'zero financial payoff' to most companies, Apollo economist says ↗

Nov 01 · 22:10

86

Deepseek v4.1 flash is now good enough to undertake huge engineering tasks autonomously ↗

Nov 05 · 11:01

85

Oracle files to sell $20B in new stock, bonds to finance AI buildout ↗

Nov 02 · 19:27

85

RefMods for MiniMax H3: consistent character + voice references, running on a 6 GB GPU (free workflow) ↗

Nov 04 · 19:36

85

Jobless Tech Workers Are Being Left Out of San Francisco's AI Boom ↗

Nov 02 · 14:40

85

Anthropic confirms new Claude Code limits to prevent game cloning ↗

Nov 02 · 18:51

84

Egoless Mathematics in the AI Era ↗

Nov 02 · 18:54

84

Minimax H3 Best Remix Tool | Friendly Ultra-Fast Character Replacement | Create 1080p Ultra-Long Videos ↗

Nov 08 · 05:11

84

Basically no one is using AI agents for their DoorDash orders ↗

Nov 02 · 15:48

83

is Opus 5.5 nerfed? ↗

Nov 08 · 00:46

83

Just a little AI resistance, as a treat ↗

Nov 03 · 14:14

83

Internal conflict at Grok. I asked it to generate examples of permitted images ↗

Nov 06 · 19:34

83

The Components of Agent Memory ↗

Nov 07 · 06:39

82

ChatGPT's search volume is now roughly 20% of Google's. Wow, people were really sick of Google, huh? ↗

Nov 04 · 05:38

82

Qualcomm Cloud AI 100 Ultra – 150W AI Accelerator Card with 128GB at 548GB/s ↗

Nov 04 · 19:55

82

I gave every Claude model the exact same prompt since July 2025. 13 models, 14 months, here's the evolution ↗

Nov 03 · 21:54

81

Launch HN: Momentic (YC W24) – Agents that catch bugs by using your app ↗

Nov 02 · 17:05

81

GPT-6 successfully performed "brain surgery" on another AI to make it smarter ↗

Nov 03 · 13:37

81

Нове OpenAI Raises $11B at a $1.25T Valuation ↗

Nov 08 · 14:49

81

Z.ai chief scientist Jie Tang just revealed the training pipeline behind the upcoming GLM-5.5 (32B, 122B, and 256B) ↗

Nov 05 · 20:06

80

China's First Frontier AI Export Controls: Mofcom Announcement No. 48 ↗

Nov 02 · 07:27

80

Viggle Remake: open weights for recasting / editing whole videos. ↗

Nov 04 · 17:38

80

Nasdaq Sinks 5% as Investors Retreat from AI Trade After Election ↗

Nov 05 · 18:00

79

Apple Watch now listens to in-person conversations & lets you rewind them ↗

Nov 04 · 17:15

79

LLMs are not superhuman at reading and writing neural network weights ↗

Nov 02 · 22:26

79

Sam Altman: "At the start of this year, AI was mostly not as smart as our smartest scientists, and now it's mostly smarter than them." He also says AI can now perform large parts of white-collar jobs that were beyond its capabilities earlier this year. ↗

Nov 02 · 17:50

79

Last night an AI agent performed brain surgery on a mouse ↗

Nov 02 · 12:50

78

Changing a character across an ~hour long TV episode ↗

Nov 06 · 21:05

78

Everything is an AI scam now ↗

Nov 04 · 21:05

78

Gemini 4 Hydrogen vs Gemini 3.8 Flash – Motion Graphics Comparison ↗

Nov 06 · 19:22

78

Agent memory needs its OpenTelemetry moment ↗

Nov 08 · 08:40

77

Super Huge Update for Minimax H3 Temporal Spatial Node 🔥🔥🔥 ↗

Nov 06 · 10:33

77

University libraries are being swarmed by AI bots, putting public access at risk ↗

Nov 06 · 14:13

77

Hideo Kojima says AI lacks what makes the creative process fun, despite its efficiency. “It’s not a form of creation I find particularly interesting” - AUTOMATON WEST ↗

Nov 05 · 19:02

76

Show HN: Uncracked City, detective game where you question LLM suspects that lie ↗

Nov 03 · 21:20

76

Linux kernel maintainers ↗

Nov 05 · 12:22

76

Launch HN: Rivet (YC F25) – Scale and Orchestrate Agents on S3 ↗

Nov 05 · 17:17

75

Democratic Arizona Congresswoman: Top of the Democratic agenda on day 1 needs to be a pause on advanced AI development, regulation of this out of control industry, and a ban on artificial superintelligence ↗

Nov 04 · 21:23

75

Launch HN: Kino AI (YC S23) – AI media manager and video editor ↗

Nov 04 · 17:31

75

Claude writes a neural network's weights by hand, without training, to detect fake legal citations—achieving 97.8% test accuracy and outperforming a conventionally trained model that scored 94.6%. ↗

Nov 04 · 14:28

75

YC's Garry Tan uses AI agents so much, he spends up to $200K a year on tokens ↗

Nov 04 · 15:28

74

Anthropic updated its S1 this morning: "we may imminently meet our definition of 'dramatically accelerated progress attributable to AI." ... "we do not anticipate being able to safely use Claude to fully autonomously design successive generations of frontier models" ↗

Nov 03 · 13:46

74

Claude Opus 5.5 Can Draw Permit-Level Architectural Drawings ↗

Nov 02 · 16:25

74

RIP Learn Mode XD ↗

Nov 02 · 01:59

74

AI for scientific discovery: How researchers use Claude ↗

Nov 06 · 09:13

73

Pro-tip: Bypassing many guardrails can be done through memory modification ↗

Nov 07 · 19:33

73

Building a Multi-Agent System from Scratch: Architecture Patterns ↗

Nov 05 · 15:51

73

Is DeepSeek V4.1 Flash getting upgrades? It built this walkable 3D yacht from one prompt for about $1.20 ↗

Nov 07 · 17:33

72

Microsoft is winding down Oracle deal to rent AI cloud capacity ↗

Nov 06 · 18:04

72

Microsoft engineer built a tool that shows what's really using your GPU ↗

Nov 07 · 15:33