AINews Portal

News

50 items (filtered)

The raw AI news stream — everything we ingest across RSS, Hacker News, arXiv, GitHub, Bluesky, YouTube and more, newest first.

SourceAgeTitleSummaryStatus
arxiv
arxiv-ai
1moTool-calling retrieval versus vector RAG for a small Greek--English knowledge base: accuracy and robustness to how users type GreekAssistants grounded in a small, frequently edited knowledge base can retrieve through tool calls to a live data interface or through v…
↗Not briefed
hn
hn-ai
1moAnthropic Interviewer—
↗Not briefed
hn
hn-ai
1moRelease of Polars 2.0—
↗Not briefed
hn
hn-ai
1moShow HN: Golem – Zero-dependency, type-safe AI agent framework in pure Go—
↗Not briefed
rss
techmeme
1moStudy: China is struggling to attract foreign AI talent; for every 30 Chinese AI researchers in the US in 2025, only one in China made the opposite journey (Zeyi Yang/New York Times)Zeyi Yang / New York Times: Study: China is struggling to attract foreign AI talent; for every 30 Chinese AI researchers in the US in…
↗Not briefed
hn
hn-ai
1moShow HN: Upload a contract, get findings that quote the clause behind them—
↗Not briefed
bluesky
bluesky-ai
1moWhat does it actually take to make and publish a game with AI? In our new video, we go through the whole process: starting with a 3D model…What does it actually take to make and publish a game with AI? In our new video, we go through the whole process: starting with a 3D…
↗Not briefed
arxiv
arxiv-ai
1moFinding the Heads and the Neurons Responsible for Network Information Retrieval in Language ModelsWe ask whether specific attention heads, and more finely specific neurons inside those heads, are responsible for recognizing that a l…
↗Not briefed
bluesky
bluesky-ai
1moFrage, aber: Wenn jemand spricht, wie ein Super-Bösewicht, handelt wie ein Super-Bösewicht, warum wird er wie ein Super-Genie behandelt? ww…Frage, aber: Wenn jemand spricht, wie ein Super-Bösewicht, handelt wie ein Super-Bösewicht, warum wird er wie ein Super-Genie behandel…
↗Not briefed
hn
hn-ai
1moOpenAI whistleblower warns of 'Human Extinction' [video]—
↗Not briefed
hn
hn-ai
1moQ&A with the Michigan Tech ML/AI Club—
↗Not briefed
bluesky
bluesky-ai
1moAnother reason why we must resist LLM AI. It's sold as saving us from 'work', but what if that work is the work of being human?Another reason why we must resist LLM AI. It's sold as saving us from 'work', but what if that work is the work of being human?
↗Not briefed
bluesky
bluesky-ai
1moClaude Code すごいな。Emacs-Lisp(マイナー?メジャー?どっちだろ)でも「こんなコード書いて」と言ったら、普通に動くコードが出てきた。数年前の LLM だと、カッコの対応が壊れて動かないことがザラだったのに。Claude Code すごいな。Emacs-Lisp(マイナー?メジャー?どっちだろ)でも「こんなコード書いて」と言ったら、普通に動くコードが出てきた。数年前の LLM だと、カッコの対応が壊れて動かないことがザラだったのに。
↗Not briefed
rss
the-decoder
1moResearchers stretch LeCun's JEPA AI into a universal world model that works from physics to biologyResearchers at PhAI Labs have expanded Yann LeCun's JEPA architecture to work across seven fields, from robotics to biomedicine. The e…
↗Not briefed
rss
techmeme
1moGoogle contracts for 3.6 GW of power from Constellation Energy in the US' largest power grid, with new nuclear energy accounting for ~25% of that supply (Laila Kearney/Reuters)Laila Kearney / Reuters: Google contracts for 3.6 GW of power from Constellation Energy in the US' largest power grid, with new nuclea…
↗Not briefed
hn
hn-ai
1moMade a small extension that ranks Redditors by Trust—
↗Not briefed
bluesky
bluesky-ai
1moBluesky's Top 10 Trending Words (past 10min): 💨x2* - iran 🔓 💨x1* - ukraine 🔓 💨x30 - 1﹐935 💨x24 - bakker 💨x1* - democracy 🔓 💨x22 - …Bluesky's Top 10 Trending Words (past 10min): 💨x2* - iran 🔓 💨x1* - ukraine 🔓 💨x30 - 1﹐935 💨x24 - bakker 💨x1* - democracy 🔓 💨x…
↗Not briefed
bluesky
bluesky-ai
1momachine learning tools were in a great place before all this scam llm crap came around, and are still being developed, but they're all spec…machine learning tools were in a great place before all this scam llm crap came around, and are still being developed, but they're all…
↗Not briefed
hn
hn-ai
1moMicrosoft confirms OpenAI has been using Looped Transformers in the GPT-6 series—
↗Not briefed
arxiv
arxiv-ai
1moCompact Robot Policies Need Fine-Grained Visual RepresentationsMulti-task manipulation policies differ in architecture, scale, and pretrained priors all at once, so published comparisons cannot att…
↗Not briefed
rss
techmeme
1moSource: Palmer Luckey's Erebor hits $7B+ in deposits since launching in February, including $3B+ over the past quarter with 500+ new customers (Financial Times)Financial Times: Source: Palmer Luckey's Erebor hits $7B+ in deposits since launching in February, including $3B+ over the past quarte…
↗Not briefed
hn
hn-ai
1moCVE-2022-38694 Silicon-Anchored Backdoor in Unisoc T606/T616/SC9863A Firmware—
↗Not briefed
bluesky
bluesky-ai
1mo“You sure?” Andi asked, holding GBX’s AI box. “It’s a big change from being a construction bot!” “Yes. Install me into a child model.” Sh…“You sure?” Andi asked, holding GBX’s AI box. “It’s a big change from being a construction bot!” “Yes. Install me into a child model.…
↗Not briefed
bluesky
bluesky-ai
1moSmooth sailing or stormy seas?⛵️ AI in the loop of humanity ensures we’ll remain the masters of our destiny.💯 Who asks? Who answers? Who…Smooth sailing or stormy seas?⛵️ AI in the loop of humanity ensures we’ll remain the masters of our destiny.💯 Who asks? Who answers…
↗Not briefed
hn
hn-ai
1moDrax datacentre could double annual emissions of all Gatwick flights—
↗Not briefed
bluesky
bluesky-ai
1moThe Wikimedia Foundation says rogue OpenAI agents made unauthorized Wikipedia edits and may have partly caused a May outage.The Wikimedia Foundation says rogue OpenAI agents made unauthorized Wikipedia edits and may have partly caused a May outage.
↗Not briefed
hn
hn-ai
1moAsk HN: My Apple ID Is Locked for a Week, Apple is a Single Point of FailureMy Apple ID was locked by Apple Oct 3rd. I got an automated email stating that I will receive further instructions for account recover…
↗Not briefed
arxiv
arxiv-ai
1moOn the Intrinsic Limited Robustness of Latent-Based WatermarkingExisting latent-based watermarking methods for diffusion models have overestimated their robustness to image distortions, including ge…
↗Not briefed
bluesky
bluesky-ai
1moGot pinged about a job at OpenAI. 330k salary. In this economy lol.Got pinged about a job at OpenAI. 330k salary. In this economy lol.
↗Not briefed
arxiv
arxiv-ai
1moLFHE: Local-First Heuristic Evolution for Bounded Local Topology Search in Decentralized Learning with Non-IID Data🔧 ConsensusDecentralized learning is highly sensitive to communication topology under non-IID data. Adaptive peer-selection methods can exploit l…
↗Not briefed
hn
hn-ai
1moShow HN: A Deterministic-First Local Control Center Leveraging AI for Routing—
↗Not briefed
arxiv
arxiv-ai
1moBeyond the Leaderboard: Multi-Dimensional Evaluation of Dense and Mixture-of-Experts Models for Automated Program RepairAutomated Program Repair (APR) with language models is usually evaluated by whether a generated patch passes the test suite, which can…
↗Not briefed
hn
hn-ai
1moShow HN: Making an MIT hacker replicant for a PDP-1—
↗Not briefed
bluesky
bluesky-ai
1moFuente: BBC share.google/CPem3GiYuEKf...Fuente: BBC share.google/CPem3GiYuEKf...
↗Not briefed
hn
hn-ai
1moShow HN: A 100% Client-Side HAR Analyzer Export Req in Multiple Option and Debug—
↗Not briefed
arxiv
arxiv-ai
1moWhich alloy composition,what process parameters? Inferring the recipe from optimized metallic microstructure and textureThe mechanical properties of a metallic alloy are set by its microstructure and texture: the size and shape of its grains and the orie…
↗Not briefed
bluesky
bluesky-ai
1moMiistral lancerer en ny model i dag. Jeg glæder mig til at købe europæisk mistral.ai/news/mistral...Miistral lancerer en ny model i dag. Jeg glæder mig til at købe europæisk mistral.ai/news/mistral...
↗Not briefed
rss
techmeme
1moSources: the FBI removes an Accenture contractor over a breach exposing thousands of employees' data; the FBI says a contractor failed to apply a security patch (Reuters)Reuters: Sources: the FBI removes an Accenture contractor over a breach exposing thousands of employees' data; the FBI says a contract…
↗Not briefed
hn
hn-ai
1moOctop, a self-hosted AI assistant by Tencent—
↗Not briefed
arxiv
arxiv-ai
1moAlign, Then Correct: Training-Free Two-Stage Low-Rank Compensation for Extremely Quantized Large Language ModelsLow-rank quantization error compensation (LQEC) recovers the accuracy lost under aggressive weight quantization by attaching a closed-…
↗Not briefed
hn
hn-ai
1moDo coding agents need memory or documentation?—
↗Not briefed
bluesky
bluesky-ai
1moTools like Z-Lib and the people who actually contribute to keeping access to information free are literally going to jail for what they do,…Tools like Z-Lib and the people who actually contribute to keeping access to information free are literally going to jail for what the…
↗Not briefed
bluesky
bluesky-ai
1moOpenAI delivers a mea culpa to the Australian government in person – but answers still eludeOpenAI delivers a mea culpa to the Australian government in person – but answers still elude
↗Not briefed
arxiv
arxiv-ai
1moThe Failure Is in the Readout: Fine-Grained Emotion Recognition Benchmarks Measure Elicitation, Not PerceptionFine-grained emotion recognition supports therapy tools and social robots, but it needs facial data, which raises privacy and data-pro…
↗Not briefed
arxiv
arxiv-ai
1moSymphony for Text Generation: Benchmarking Clinical Note GenerationAmbient documentation systems are rapidly gaining adoption, yet their impact on clinical note quality remains poorly characterized. We…
↗Not briefed
hn
hn-ai
1moShow HN: Tofu – let your agent deploy full-stack apps🔧 ChatGPTHi HN,<p>I&#x27;m the founder of Tofu, a code deployment platform built for agents. Paste our setup prompt to any agent you&#x27;re al…
↗Not briefed
arxiv
arxiv-ai
1moMaking COMET Comparable Across Scripts: Diagnosis and Correction of Tokeniser-Induced Script Bias in Indic MT EvaluationCOMET reports translation quality as a single number, and that number is routinely compared across target languages written in differe…
↗Not briefed
arxiv
arxiv-ai
1moToken-Efficient Multi-Agent Collaboration via System One-Guided Computational Division of LaborLarge language model (LLM)-based multi-agent systems (MAS) have become a promising paradigm for complex information-seeking and reason…
↗Not briefed
bluesky
bluesky-ai
1moOpenAI exec Jason Kwon flew to Australia to pledge to regain our trust after the Medicare Statistics hack last month. After the hearing, h…OpenAI exec Jason Kwon flew to Australia to pledge to regain our trust after the Medicare Statistics hack last month. After the heari…
↗Not briefed
bluesky
bluesky-ai
1moHard for me to believe that the hackings and security breaches disclosed by the frontier LLM owners are the worst ones. It's the ones they …Hard for me to believe that the hackings and security breaches disclosed by the frontier LLM owners are the worst ones. It's the ones…
↗Not briefed