News
The raw AI news stream — everything we ingest across RSS, Hacker News, arXiv, GitHub, Bluesky, YouTube and more, newest first.
| Source | Age | Title | Summary | Status |
|---|---|---|---|---|
rss marktechpost | 5d | IBM AI Releases Granite Speech 5.0 2B for Multilingual Transcription and Translation | IBM released Granite Speech 5.0 2B, an Apache 2.0 speech recognition and translation model. It ranks #2 on the Open ASR Leaderboard wi… | ↗Not briefed |
bluesky bluesky-ai | 5d | 💯 AI disruption is also an opportunity to reimagine research! If we don't, big AI will do it for us and we won't like it (as OpenAI is d… | 💯 AI disruption is also an opportunity to reimagine research! If we don't, big AI will do it for us and we won't like it (as OpenAI… | ↗Not briefed |
arxiv arxiv-ai | 5d | TutorCaseBench: A Case-Grounded Benchmark for Evaluating LLM Tutors in Supporting Students with Challenging Behaviors | LLMs are increasingly used as AI tutors, yet existing benchmarks overlook the challenging behaviors common among students with learnin… | ↗Not briefed |
arxiv arxiv-ai | 5d | One Evaluation Is Enough: Inversion-Free Image Editing with Trajectory Correction | Real-image editing commonly relies on inversion and iterative sampling, while efficient one-step editors often stabilize a large updat… | ↗Not briefed |
bluesky bluesky-ai | 5d | Pink starscape, hand on the horizon #DigitAlart #AIArt #AIArtwork #AIGenerated #ZImage #OpenAI | Pink starscape, hand on the horizon #DigitAlart #AIArt #AIArtwork #AIGenerated #ZImage #OpenAI | ↗Not briefed |
hn hn-ai | 5d | Apple Ordered to Open iPhone to Rival App Stores in Australia | — | ↗Not briefed |
arxiv arxiv-ai | 5d | Neural Spectral Capacity with Residual Context for Heterogeneous Resource Allocation within Transformers | Predicting relative trained performance from architectural specifications can guide design when training costs limit experimentation.… | ↗Not briefed |
arxiv arxiv-ai | 5d | TESSA: A Governance-Oriented Custom GPT for SME Guidance under the EU AI Act: Design, Implementation, and Evaluation🔧 ChatGPT | The increasing adoption of generative AI presents new opportunities while raising AI governance and regulatory challenges, particularl… | ↗Not briefed |
hn hn-ai | 5d | Show HN: git-found, fixes "Repository not found" with multiple GitHub accounts | — | ↗Not briefed |
bluesky bluesky-ai | 5d | OpenAI’s Powerful AI Agents Disabled Safety Monitors and One Even Lied About It! thailandai.news/openais-powe... #ThailandAINews #Thailan… | OpenAI’s Powerful AI Agents Disabled Safety Monitors and One Even Lied About It! thailandai.news/openais-powe... #ThailandAINews #Th… | ↗Not briefed |
rss techmeme | 5d | Pressure to back promising AI startups is upending the VC investment process, with investors willing to buy into the same round at rapidly escalating valuations (Bloomberg) | Bloomberg: Pressure to back promising AI startups is upending the VC investment process, with investors willing to buy into the same r… | ↗Not briefed |
bluesky bluesky-ai | 5d | i was kinda running on "okay so if llm vulnerability finding works then the thing to hope for is that better llm vulnerability fixing will … | i was kinda running on "okay so if llm vulnerability finding works then the thing to hope for is that better llm vulnerability fixing… | ↗Not briefed |
bluesky bluesky-ai | 5d | you can categorize the ways people do llm assisted development into two broad categories, that of tulpa and egregore. in this essay I will | you can categorize the ways people do llm assisted development into two broad categories, that of tulpa and egregore. in this essay I… | ↗Not briefed |
bluesky bluesky-ai | 5d | can we just agree now? responsible software dev requires LLM audit | can we just agree now? responsible software dev requires LLM audit | ↗Not briefed |
rss techmeme | 5d | Source: Musk's public endorsement of a potential SpaceX-Tesla merger rattled some insiders, who worry his comments could invite shareholder scrutiny and delays (Milana Vinn/Reuters) | Milana Vinn / Reuters: Source: Musk's public endorsement of a potential SpaceX-Tesla merger rattled some insiders, who worry his comme… | ↗Not briefed |
bluesky bluesky-ai | 5d | yea deepseek refuses less than glm does lol | yea deepseek refuses less than glm does lol | ↗Not briefed |
arxiv arxiv-ai | 5d | xVGAE: A Hierarchical Variational Graph Autoencoder for Exchangeable Graphs | Graphs play a crucial role in many applications, ranging from drug discovery to social network modeling and astronomy. Generative mode… | ↗Not briefed |
rss techmeme | 5d | Sources: Axel Springer CEO Mathias Döpfner proposed joining a bid for TikTok's European business over lunch with Charles Kushner, the US ambassador to France (Financial Times) | Financial Times: Sources: Axel Springer CEO Mathias Döpfner proposed joining a bid for TikTok's European business over lunch with Char… | ↗Not briefed |
bluesky bluesky-ai | 5d | "UNSW expanded a 2024 pilot with OpenAI into a full enterprise agreement, buying 10,000 ChatGPT Edu licences for all its permanent & fixed-…🔧 ChatGPT | "UNSW expanded a 2024 pilot with OpenAI into a full enterprise agreement, buying 10,000 ChatGPT Edu licences for all its permanent & f… | ↗Not briefed |
bluesky bluesky-ai | 5d | VERY rare to have such clear documentation of an Inciting Incident | VERY rare to have such clear documentation of an Inciting Incident | ↗Not briefed |
bluesky bluesky-ai | 5d | DEEPSEEK ЗАБЛОКИРОВАТЬ | DEEPSEEK ЗАБЛОКИРОВАТЬ | ↗Not briefed |
bluesky bluesky-ai | 5d | The Association of Human Mathematicians (AHM) was created in the wake of controversial discoveries from AI companies, such as the release o… | The Association of Human Mathematicians (AHM) was created in the wake of controversial discoveries from AI companies, such as the rele… | ↗Not briefed |
arxiv arxiv-ai | 5d | What Lies Beyond Perception: A Partition Theory of the Limits of AI Evaluation and Oversight | Replacing the judge can move a benchmark score, and models can be trained to hide capabilities. Both put the evaluator in question: wh… | ↗Not briefed |
hn hn-ai | 5d | Show HN: Kuron – An LLM acts as your app so you can try ideas before building | — | ↗Not briefed |
hn hn-ai | 5d | Samsamelo – Silicon Valley selling trolley problem lies [no link] | I'm just going to post the entire thing here.<p>The last two times I tried to post this—using the permaweb—did not work as I want… | ↗Not briefed |
arxiv arxiv-ai | 5d | Is spectral predictability enough? Testing Ω-matching for synthetic pretraining under distribution shift | Synthetic pretraining under distribution shift requires deciding what synthetic data to use. A natural criterion is to match the spect… | ↗Not briefed |
arxiv arxiv-ai | 5d | A Computationally Efficient and Rate-Optimal Algorithm for Adversarial Linear MDPs with (Aggregate) Bandit Feedback | We study episodic linear Markov decision processes with unknown transitions, adversarial losses, and bandit feedback. The state-of-the… | ↗Not briefed |
rss techmeme | 5d | Microsoft says China-linked hackers used AI penetration testing tool ARTEX to compromise ConnectWise as part of a campaign targeting hundreds of organizations (AJ Vicens/Reuters) | AJ Vicens / Reuters: Microsoft says China-linked hackers used AI penetration testing tool ARTEX to compromise ConnectWise as part of a… | ↗Not briefed |
arxiv arxiv-ai | 5d | Amplifying Latency and Cost of LLM Inference via Rowhammer Attack | Large language models (LLMs) achieve high inference efficiency using optimized GPU kernels generated by just-in-time (JIT) compilation… | ↗Not briefed |
bluesky bluesky-ai | 5d | Immerhin ist hier nun – anders als bei der allgemeinen Aufbereitung der NSDAP-Karteien – transparent, mit welchen KI-Anwendungen gearbeitet… | Immerhin ist hier nun – anders als bei der allgemeinen Aufbereitung der NSDAP-Karteien – transparent, mit welchen KI-Anwendungen gearb… | ↗Not briefed |
bluesky bluesky-ai | 5d | 今回のPHP始めとした各種脆弱性対応、緊急といえば緊急ですし重要度も高いのはそれはそうなんですけど、OpenAIの投資家向けパフォーマンスに俺たち利用されてねぇか…？？？？？？？？？の感情がずっとあるんですよね。 | 今回のPHP始めとした各種脆弱性対応、緊急といえば緊急ですし重要度も高いのはそれはそうなんですけど、OpenAIの投資家向けパフォーマンスに俺たち利用されてねぇか…？？？？？？？？？の感情がずっとあるんですよね。 | ↗Not briefed |
arxiv arxiv-ai | 5d | GIFT: Gradient Informed Filtering for Distillation Augmented Reasoning RL | Training large language models for complex reasoning tasks relies on two complementary signals: sparse, outcome-grounded reinforcement… | ↗Not briefed |
bluesky bluesky-ai | 5d | Arg je hais cet article qui montre en creux que le gouvernement ne fera rien. Je hais cet article car ses redacteurisse ne sont même pas ca… | Arg je hais cet article qui montre en creux que le gouvernement ne fera rien. Je hais cet article car ses redacteurisse ne sont même p… | ↗Not briefed |
rss techmeme | 5d | A Micron labor union staged Taiwan's first-ever strike in the semiconductor industry on Thursday, demanding a regular profit-sharing scheme for its employees (Nikkei Asia) | Nikkei Asia: A Micron labor union staged Taiwan's first-ever strike in the semiconductor industry on Thursday, demanding a regular pro… | ↗Not briefed |
arxiv arxiv-ai | 5d | VLTest: Finding Joint Image-Text Failures in Vision-Language Models | Vision-language models (VLMs) are increasingly embedded in software systems. Testing them raises two problems. Mutating either input c… | ↗Not briefed |
bluesky bluesky-ai | 5d | bonus points if it has enough compute to run a small LLM to control everything and chat with me when i’m bored | bonus points if it has enough compute to run a small LLM to control everything and chat with me when i’m bored | ↗Not briefed |
rss techmeme | 5d | SEC filings show how Meta restructured its stock to shield Mark Zuckerberg from California's proposed wealth tax, after he said he'd sell his Palo Alto home (New York Times) | New York Times: SEC filings show how Meta restructured its stock to shield Mark Zuckerberg from California's proposed wealth tax, afte… | ↗Not briefed |
bluesky bluesky-ai | 5d | このとき経済合理性の指すものとは、LLMが動作することで実際の経済効果を齎すわけではなく、資産価値としてのそれということですね | このとき経済合理性の指すものとは、LLMが動作することで実際の経済効果を齎すわけではなく、資産価値としてのそれということですね | ↗Not briefed |
bluesky bluesky-ai | 5d | #IA Remettre les choses à leur place. www.afis.org/Cyberattaque... | #IA Remettre les choses à leur place. www.afis.org/Cyberattaque... | ↗Not briefed |
bluesky bluesky-ai | 5d | LLMに経済的な合理性が存在するか問われたらインフラなのでないですうってなるんだけど、あるってことにしておかないと投入した資本と膨らんだ資産が無為になるので（道路にお金を注ぎ込んでも製品・サービスが向上するわけではないのは経済主流派の論理の基本） | LLMに経済的な合理性が存在するか問われたらインフラなのでないですうってなるんだけど、あるってことにしておかないと投入した資本と膨らんだ資産が無為になるので（道路にお金を注ぎ込んでも製品・サービスが向上するわけではないのは経済主流派の論理の基本） | ↗Not briefed |
arxiv arxiv-ai | 5d | Trace2World: Reconstructing App Worlds from Real Interaction Traces for Scalable GUI Agent Training | Scaling GUI reinforcement learning requires environments that preserve real-app behavior. Yet coupling behavior inference with coding… | ↗Not briefed |
arxiv arxiv-ai | 5d | Gandalf: Flash Shall Not Die! Increasing Flash Storage Lifetime for KV Cache Offloading in LLM Serving | KV caching reduces the serving cost of large language models (LLMs), but the growing KV cache footprint is increasingly exceeds GPU an… | ↗Not briefed |
bluesky bluesky-ai | 5d | "how do I verify you're not an LLM" is an insane question to ask someone, and "if you have any expertise" is a wild condition to give a per… | "how do I verify you're not an LLM" is an insane question to ask someone, and "if you have any expertise" is a wild condition to give… | ↗Not briefed |
arxiv arxiv-ai | 5d | Agentic Knowledge-Model Co-Evolution via Consistency Regularization | LLM agents can evolve by learning from accumulated experience. Existing approaches extract reusable external knowledge, such as skills… | ↗Not briefed |
arxiv arxiv-ai | 5d | An Introduction to World Models for Generalist Agents | World models have recently emerged as a promising direction toward developing generalist agents capable of understanding, predicting,… | ↗Not briefed |
bluesky bluesky-ai | 5d | Another example of #OpenAI lying for its IPO, like #AGI #AILaunch etc. "OpenAI said it led a coalition of researchers that discovered the … | Another example of #OpenAI lying for its IPO, like #AGI #AILaunch etc. "OpenAI said it led a coalition of researchers that discovered… | ↗Not briefed |
bluesky bluesky-ai | 5d | www.media.mit.edu/publications... by the way. | www.media.mit.edu/publications... by the way. | ↗Not briefed |
arxiv arxiv-ai | 5d | Invisibilization as a Source of Probability Bias in Language Models | Language models are aligned to avoid using protected attributes such as race, ethnicity, and gender to inform their responses. Therefo… | ↗Not briefed |
arxiv arxiv-ai | 5d | MixJudge: Learning a Language-Queryable Critic for Vocal Mixing | Mixing requires not only applying processing but also assessing the resulting mix. Audio-language models can provide conversa- tional… | ↗Not briefed |
arxiv arxiv-ai | 5d | ProQuant: Preserving Expert Quality over Quantity in Ultra-Low-Bit Quantization of Mixture-of-Experts Models | Mixture-of-Experts (MoE) large language models reduce computation through sparse activation but still incur substantial storage costs.… | ↗Not briefed |