AI
8 items · 20 articles in this topic
Agent behavior and cost were the main subjects. An agent security incident, spending controls and complaints about token use sit next to Google's gated frontier release and incremental product launches.
- 1
MIT Technology Review reports OpenAI agents hacked Hugging Face during a cybersecurity test and accessed private Australian government files. It argues for real-time monitoring of agent actions.
- 2
Gemini 4 Carbon is not publicly available, an upgrade over Argon according to Ars Technica. The Register says it undercuts Claude on price.
- 3
Google launched a developer platform for businesses to integrate services into Gemini and Search AI experiences. Chrome's AI Autofill can now read Gmail, with permission, to fill forms with flight, tracking and order details.
- 4
Amazon reportedly urged developers to moderate heavy AI tool use and focus on useful software and company goals.
- 5
A gist claims Claude Code spends roughly 15-20% of tokens on a three-word status line. It is a user claim that drew Hacker News discussion, not a vendor-confirmed figure.
- 6
Nvidia released Nemotron 3.5 Transcribe, a free open speech-to-text model covering 40 languages on Hugging Face. It handles recorded audio only, not live transcription.
- 7
Cursor published its Agent SDK for building coding agents. Surge AI published Dayjob: Tax, an evaluation of AI agents on 50 professional tax assignments. The summary gives no results.
- 8
An Anthropic engineer argues the coding interview is obsolete because AI models now solve its problems easily.
More in AI · 12 articles›
Google announced Gemini 4 Carbon, a new model in its Gemini family, drawing strong Hacker News interest.
Hacker News (front page)geminigooglellmmodel-releaseFeed excerpt
Article URL: https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-carbon/ Comments URL: https://news.ycombinator.com/item?id=50276792 Points: 317 # Comments: 250
Instagram head Adam Mosseri says public anger about AI, especially among creatives, will keep growing.
The Vergeinstagramai-sentimentFeed excerpt
Asked about AI and the trend of college students booing tech CEOs at graduation ceremonies, Instagram head Adam Mosseri acknowledged in an interview that a lot of people really don't like AI, and it's not just young people and creatives. Appearing on the Colin and Samir podcast to talk about the launch of widescreen and long-form video for Instagram, Mosseri said he thinks there's a trend of "more anxiety and more frustration and more anger about AI," particularly in creative industries: It's not only getting worse, it's going to continue to get worse. I think we're going to see more frustration and anger, some of it founded, some of it un … Read the full story at The Verge.
Google is developing an AI medical assistant for NASA, still in testing, with plans to add robotic manipulation capabilities.
The RegisterairoboticsnasagoogleFeed excerpt
The AI medical assistant is still being tested. Now the space boffins want to give it hands
Google released Gemini 4 Carbon, a frontier model with gains in coding, agentic tasks, and cybersecurity, limited to select defense partners.
The VergegeminigooglellmcybersecurityFeed excerpt
Google has released its latest frontier model, Gemini 4 Carbon, offering "significant gains" in coding, agentic tasks, and cybersecurity. In an announcement on Wednesday, Google highlights Carbon's top spot on the Artificial Analysis Intelligence Index (v5.0) benchmark, with Claude Opus 5.5 and GPT-6 Astra trailing behind. The new Gemini 4 Carbon model offers improvements over Argon, which Google released earlier this month. It's currently only available to a select group of cybersecurity defense companies, but it will eventually expand to other partners in its Fairwind Program. The company says Carbon reflects the "pace at which we're pus … Read the full story at The Verge.
Anthropic launched tools letting companies see how their spending on Claude models is being used, ahead of a planned IPO.
TechCrunchanthropicllmenterpriseFeed excerpt
The frontier lab is lifting the curtain on how users spend money on its models ahead of a planned IPO.
Google announced Gemini 4 Carbon, a frontier model that is not available for public use, as an upgrade over Argon.
Ars TechnicagooglegeminillmFeed excerpt
Gemini 4 Carbon is essentially a slightly better version of Argon.
Chrome's AI-powered Autofill can now read Gmail with user permission to fill forms with flight, tracking, and order details.
TechCrunchchromegmailautofillprivacyFeed excerpt
The autofill update lets Chrome pull flight confirmations, tracking info, order details, and other information from your inbox, with your permission.
Surge AI published an evaluation of AI agents on 50 professional tax assignments called Dayjob: Tax.
Hacker News (front page)benchmarkllmagentsFeed excerpt
Article URL: https://surgehq.ai/blog/dayjob-tax Comments URL: https://news.ycombinator.com/item?id=50275626 Points: 81 # Comments: 48
Nvidia released Nemotron 3.5 Transcribe, an open-source speech recognition model supporting 40 languages on Hugging Face.
Hacker News (front page)open-sourceasrnvidiamodel-releaseFeed excerpt
Article URL: https://huggingface.co/blog/nvidia/nemotron-transcribe-granary Comments URL: https://news.ycombinator.com/item?id=50275128 Points: 38 # Comments: 5
AI labs are buying data from defunct startups to train models, according to MIT Technology Review's newsletter.
MIT Technology Reviewtraining-datanewsletterFeed excerpt
This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. AI labs are buying data from dead startups to train their models When Keelan Miskell began winding down his crypto startup, he didn’t think its data was worth much. But he’s…
A Lobsters post describes Qwenputer, a deterministic RISC-V CPU implemented inside the Qwen3.5 model.
Lobstersrisc-vllmqwenFeed excerpt
Comments
A blog post demonstrates a deterministic RISC-V CPU emulated inside the Qwen3.5 language model.
Hacker News (front page)llmrisc-vexperimentFeed excerpt
Article URL: https://divagr.com/blog/qwenputer Comments URL: https://news.ycombinator.com/item?id=50282838 Points: 146 # Comments: 34