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
hn
hn-ai
16dWaveDude Next: 2D interactive 2D wave optics simulator—
↗Not briefed
rss
the-decoder
16dNo AI agent scores above 12 percent on SemiAnalysis' new real-world work benchmarkSemiAnalysis is putting AI agents through simulated knowledge work in its new GDPx benchmark, covering everything from chip design to…
↗Not briefed
arxiv
arxiv-ai
16dLeanContext: Disentangling Mathematical Resources from Local Formal Context in Lean Whole-Proof GenerationNeural theorem proving increasingly uses retrieved premises and project context to support proof generation. Evaluating these resource…
↗Not briefed
hn
hn-ai
16dSoftware Line Specialist—
↗Not briefed
arxiv
arxiv-ai
16dAligning Policy Encoders with a Shape Teacher3D-aware representations can improve robotic manipulation by providing geometric information that is robust to changes in viewpoint, o…
↗Not briefed
arxiv
arxiv-ai
16dAccelerating Safe Reinforcement Learning with Massive ParallelismWe investigate the problem of scaling safe reinforcement learning (safe RL) to massively parallel training regimes, where a large numb…
↗Not briefed
bluesky
bluesky-ai
16dllm protein modeling is cool but we can safely turn off the picture machines. the drawbacks are not worth images of amy rose dressed like 2…llm protein modeling is cool but we can safely turn off the picture machines. the drawbacks are not worth images of amy rose dressed l…
↗Not briefed
arxiv
arxiv-ai
16dDART: Domain-Agnostic Residual Transfer for Generalist Anomaly DetectionGeneralist Anomaly Detection (GAD) aims to detect anomalies in unseen domains using only a few reference samples. Recent normal-abnorm…
↗Not briefed
hn
hn-ai
16dParaguay seizes major digital bank tied to president, leaving customers in limbo—
↗Not briefed
hn
hn-ai
16dTick – A Smalltalk with a Tcl like syntax—
↗Not briefed
arxiv
arxiv-ai
16dAccepted but Wrong: Silent Failures of LLM Code Repairs after API Migrations🔧 GPT-5After a runtime or API migration, an LLM-repaired procedure can run cleanly yet return wrong results, and a release check on typical t…
↗Not briefed
rss
techmeme
16dSources: Microsoft is tightening oversight of some data center projects, stepping away from those it deems risky, and debating whether to ink new neocloud deals (Brody Ford/Bloomberg)Brody Ford / Bloomberg: Sources: Microsoft is tightening oversight of some data center projects, stepping away from those it deems ris…
↗Not briefed
hn
hn-ai
16dDiscord Plans Lighter Desktop App with Rust and GPUI—
↗Not briefed
bluesky
bluesky-ai
16d...a billion AI slop images and not a single one is the thing the person imagined when they started to mash the keyboard... ...not a singl…...a billion AI slop images and not a single one is the thing the person imagined when they started to mash the keyboard... ...not a…
↗Not briefed
arxiv
arxiv-ai
16dScaling Neural Motor Decoding via Decoupled Behavioral PretrainingRecent neural decoding approaches achieve strong performance by training on large collections of paired neural-behavioral recordings,…
↗Not briefed
arxiv
arxiv-ai
16dThe Geometry of Simulated Perspectives: Dissociating Representation and Behaviour in a Multimodal Language ModelMultimodal language models can simulate different observers, yet it remains unclear how assigned perspectives reshape their internal r…
↗Not briefed
rss
techmeme
16dGreece's parliament approves a 10% capital gains tax on crypto, exempting gains of up to €500 a year, in the country's first comprehensive crypto tax framework (Lefteris Papadimas/Reuters)Lefteris Papadimas / Reuters: Greece's parliament approves a 10% capital gains tax on crypto, exempting gains of up to €500 a year, in…
↗Not briefed
hn
hn-ai
16dTobacco company got smoked by leaving secret recipes on a file share—
↗Not briefed
bluesky
bluesky-ai
16dOn AI efficiency gains: "It's bullish for the labs and terribly bearish for everyone that does data centers. Now, that could be an overlap…On AI efficiency gains: "It's bullish for the labs and terribly bearish for everyone that does data centers. Now, that could be an ov…
↗Not briefed
bluesky
bluesky-ai
16dPerfect timing again by Premier Tim Houston. Forcing an Anthropic data center on us at a time when they're set to crash, just like the "Gre…Perfect timing again by Premier Tim Houston. Forcing an Anthropic data center on us at a time when they're set to crash, just like the…
↗Not briefed
arxiv
arxiv-ai
16dNowcasting of Tropical Cyclone Wind Fields and Intensity from Geostationary ImageryHigh-resolution tropical cyclone surface wind speeds derived from synthetic aperture radar (SAR) observations can support storm struct…
↗Not briefed
bluesky
bluesky-ai
16dIt's worth remembering that Ro is a mixed bag, but his district encompasses much of Silicon Valley, so he's effectively the de facto Tech R…It's worth remembering that Ro is a mixed bag, but his district encompasses much of Silicon Valley, so he's effectively the de facto T…
↗Not briefed
hn
hn-ai
16dThe AI Times – wake up to a printed paper, not a feed—
↗Not briefed
arxiv
arxiv-ai
16dSeries-wise Cross-validation for Global Time Series ForecastingWe consider the problem of cross-validation (CV) in time series forecasting, with a focus on global forecasting models based on deep l…
↗Not briefed
rss
wired-ai
16dReddit Founder Says Building Social Media Apps Is ‘Not Cool Anymore’Alexis Ohanian is trying to figure out what’s next for user-generated content now that so much of the internet is “compromised.”
↗Not briefed
arxiv
arxiv-ai
16dGeometric Relevance Is Not Sufficient: Evaluating Focused Post-Training Supervision in Diffusion DockingDiffusion-based molecular docking requires learning a corrective score field across a broad corruption distribution, yet individual po…
↗Not briefed
bluesky
bluesky-ai
16dLittle nugget in this piece. Dan Driscoll, former secretary of the army who resigned in August, has been hired by OpenAi.Little nugget in this piece. Dan Driscoll, former secretary of the army who resigned in August, has been hired by OpenAi.
↗Not briefed
hn
hn-ai
16dGoogle Summaries Aren't About AI, They're About Lock-In—
↗Not briefed
rss
techmeme
16dGoogle, Meta, and TikTok file lawsuits to block a California law that bans online platforms from providing "psychologically exploitative features" to under-16s (Jonathan Stempel/Reuters)Jonathan Stempel / Reuters: Google, Meta, and TikTok file lawsuits to block a California law that bans online platforms from providing…
↗Not briefed
hn
hn-ai
16dRemote Desktop for Mac in 2026—
↗Not briefed
bluesky
bluesky-ai
16dGod it's so good. Everywhere I look in the LLM attached markets people are taking fat Ls. Also got to read that story about Apple reducing …God it's so good. Everywhere I look in the LLM attached markets people are taking fat Ls. Also got to read that story about Apple redu…
↗Not briefed
bluesky
bluesky-ai
16dStill seeing indie devs spending days hand-crafting influence maps and calling basic conditional logic "AI". Real progress is hooking LLM …Still seeing indie devs spending days hand-crafting influence maps and calling basic conditional logic "AI". Real progress is hooking…
↗Not briefed
bluesky
bluesky-ai
16dIt’s like Uber, which has never made a profit. It survives on constant new investment. Its (stated, published) plan is to push competition …It’s like Uber, which has never made a profit. It survives on constant new investment. Its (stated, published) plan is to push competi…
↗Not briefed
arxiv
arxiv-ai
16dValue-oriented policy distillation for interpretable wind farm power controlWind farm power control can help the system maintain balance and avoid blackouts. In Europe, this can be achieved through implicit bal…
↗Not briefed
bluesky
bluesky-ai
16dCommandeer a prompt and it still gets answered by whichever model the router picked. Routing shapes what users feel and costs ops hours. A …Commandeer a prompt and it still gets answered by whichever model the router picked. Routing shapes what users feel and costs ops hour…
↗Not briefed
hn
hn-ai
16dAsk HN: Local AI Models for Coding and General Tasks?Hi,<p>I&#x27;m looking at Gemma and Qwen as a local AI. I wanted to try it out for coding but also some other basic tasks. My system h…
↗Not briefed
hn
hn-ai
16dMy psychophysics as a bilateral cochlear implant user—
↗Not briefed
hn
hn-ai
16dShow HN: BeatBox – a cybersecurity lab on your own computer (QEMU, Rust)—
↗Not briefed
arxiv
arxiv-ai
16dFewStepRL: Fine-tuning Few-step Diffusion Models with Reinforcement LearningReinforcement learning (RL) has become a standard post-training method for diffusion models, but most RL objectives roll out the curre…
↗Not briefed
arxiv
arxiv-ai
16dTerminal-style Mathematical Modeling Bench: The First Agentic and Verifiable Benchmark for Mathematical ModelingMathematical modeling turns an open-ended real-world problem into assumptions, a model, a solution, and a defended recommendation, and…
↗Not briefed
arxiv
arxiv-ai
16dBeyond Quadratic Attention: Separable Gaussian Operators for ODE Vision TransformersSelf-attention carries a quadratic cost in token count, which limits how far Vision Transformers can scale with input resolution. We i…
↗Not briefed
bluesky
bluesky-ai
16d2/ At last year’s remedies hearing, Google testified that letting publishers opt out of its AI products one by one would be hard to support…2/ At last year’s remedies hearing, Google testified that letting publishers opt out of its AI products one by one would be hard to su…
↗Not briefed
rss
techcrunch-ai
16dAxiom Math wants to keep AI slop out of mathematicsAs AI floods math with unchecked proofs, Axiom Math is betting on verification. The startup’s AI agents just formally verified the Sta…
↗Not briefed
arxiv
arxiv-ai
16dSignals, Evidence, and Context: User Scrutiny in Real-World LLM ConversationsLarge language models (LLMs) often present synthesized answers without making their supporting evidence easy to inspect. We study how…
↗Not briefed
arxiv
arxiv-ai
16dFrom Twisted Facts to Rogue Agents: Can LLMs Reliably Detect Multi-Dimensional Hallucinations?Large language models (LLMs) are widely used for information access, content production, and agent-based task execution, but their out…
↗Not briefed
bluesky
bluesky-ai
16dOn Anthropic's model efficiency: "The most likely outcome ... is that, if there really is an ultra-efficient way of making these models .…On Anthropic's model efficiency: "The most likely outcome ... is that, if there really is an ultra-efficient way of making these mod…
↗Not briefed
hn
hn-ai
16dClaude Study – circle part of your handwritten notes and ask Claude—
↗Not briefed
bluesky
bluesky-ai
16dThis could be the best news of the century (and I say this the day after the very welcome Midterms, please note...) So, why are you keepin…This could be the best news of the century (and I say this the day after the very welcome Midterms, please note...) So, why are you k…
↗Not briefed
arxiv
arxiv-ai
16d"Blind Compliance": Coercive Control in AI JailbreakingThe explosion of generative AI systems and concerns surrounding their rapidly scaling capabilities has led to an adversarial cycle bet…
↗Not briefed
rss
techmeme
16dMotorola launches the $1,599+ Signature 27, featuring a Snapdragon 8 Elite Extreme Gen 6 chipset, a 6.8-inch Tandem OLED display, and seven years of OS updates (Ben Schoon/9to5Google)Ben Schoon / 9to5Google: Motorola launches the $1,599+ Signature 27, featuring a Snapdragon 8 Elite Extreme Gen 6 chipset, a 6.8-inch…
↗Not briefed