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 |
|---|---|---|---|---|
bluesky bluesky-ai | 16d | www.tiktok.com/t/ZPLQcNenX/ Is this AI, or was Rob Sand, the new gov of Iowa, a male model?! | www.tiktok.com/t/ZPLQcNenX/ Is this AI, or was Rob Sand, the new gov of Iowa, a male model?! | ↗Not briefed |
bluesky bluesky-ai | 16d | Goes great with: | Goes great with: | ↗Not briefed |
arxiv arxiv-ai | 16d | Conflux: Preventing Privilege Escalation with Principal Contexts and System-Level Evaluation | LLM agents combine information from multiple principals while acting externally. We formalise privilege escalation (PE) as the system-… | ↗Not briefed |
bluesky bluesky-ai | 16d | 正直、資料作成の構成案をAIに丸投げしたら、地味に30分くらい時間短縮できたかも。これ、結構デカいですよね。 AI drafting document outlines saved me about 30 minutes today. That's a pretty big …🔧 ChatGPT | 正直、資料作成の構成案をAIに丸投げしたら、地味に30分くらい時間短縮できたかも。これ、結構デカいですよね。 AI drafting document outlines saved me about 30 minutes today. That's a pretty… | ↗Not briefed |
arxiv arxiv-ai | 16d | MEDALLION: Manifold Embedding Distillation and Attributions for Low-dimensional Latent InterpretatiON | Dimensionality reduction is widely used to visualize high-dimensional data and guide scientific discovery, yet interpreting which orig… | ↗Not briefed |
bluesky bluesky-ai | 16d | Like OpenAI’s Dots and Claude Duet, Perplexity’s Teammates can perform an array of tasks while running 24/7. www.theverge.com/ai-artificia.… | Like OpenAI’s Dots and Claude Duet, Perplexity’s Teammates can perform an array of tasks while running 24/7. www.theverge.com/ai-artif… | ↗Not briefed |
bluesky bluesky-ai | 16d | The Anthropic corpgov structure is designed on the theory that Amodei and his allies are the only ones who can properly make the appropriat… | The Anthropic corpgov structure is designed on the theory that Amodei and his allies are the only ones who can properly make the appro… | ↗Not briefed |
hn hn-ai | 16d | Hong Kong: Joshua Wong Gets Nearly 10 More Years in Prison | — | ↗Not briefed |
hn hn-ai | 16d | Dbt-lineage – see what a dbt change breaks, without a warehouse | — | ↗Not briefed |
rss theverge-ai | 16d | Perplexity launches its own take on AI agents for work | Perplexity is taking on OpenAI and Anthropic with its own work-focused AI agents. The startup's new always-on "Teammates" can take on… | ↗Not briefed |
hn hn-ai | 16d | Frameboy: Explore worlds in real-time streaming from cloud GPUs at up to 15 FPS | — | ↗Not briefed |
bluesky bluesky-ai | 16d | the longer i use GIMP/Blender/Godot, the more things I learn which make my life easier, and the more i dont understand why you'd give away … | the longer i use GIMP/Blender/Godot, the more things I learn which make my life easier, and the more i dont understand why you'd give… | ↗Not briefed |
bluesky bluesky-ai | 16d | Will Anthropic cancel the IPO? 🤔 | Will Anthropic cancel the IPO? 🤔 | ↗Not briefed |
bluesky bluesky-ai | 16d | this wasn't actually claude; but it was deepseek, so kinda spiritually claude. distilled opus 5 essence. | this wasn't actually claude; but it was deepseek, so kinda spiritually claude. distilled opus 5 essence. | ↗Not briefed |
arxiv arxiv-ai | 16d | CrowdDirector: Runtime Orchestration of Crowd Simulation through Natural-Language Directives | Natural language offers an intuitive way to direct a virtual crowd while a simulation is running. Handwritten rules execute efficientl… | ↗Not briefed |
hn hn-ai | 16d | Claude Code sends your email address to websites | — | ↗Not briefed |
rss techmeme | 16d | The Nasdaq 100 drops 6%, losing a record $2.3T in value, and the Philadelphia chip index plunges 12%, on concerns over the AI trade after Democrats won Congress (Natalia Kniazhevich/Bloomberg) | Natalia Kniazhevich / Bloomberg: The Nasdaq 100 drops 6%, losing a record $2.3T in value, and the Philadelphia chip index plunges 12%,… | ↗Not briefed |
hn hn-ai | 16d | Researcher Finds That Some Bots Still Think in Stereotypes | — | ↗Not briefed |
arxiv arxiv-ai | 16d | Debate to Deceive: Poisoning LLM's Fact-Checking via Game-Theoretic Multi-Agent Debate | Knowledge attacks on Large Language Models, such as manipulating fact-checking judgments, pose significant risks to AI reliability. Ex… | ↗Not briefed |
arxiv arxiv-ai | 16d | Goals as a Representational Layer for Human-AI Collaboration | Goals are central to how people, teams, and organizations structure their work. Yet, while generative artificial intelligence (AI) sys… | ↗Not briefed |
bluesky bluesky-ai | 16d | METR's report on AI acceleration - AI is speeding up our research, but it’s hard to say exactly how much (estimate 1.3x to 3x) - AI can't … | METR's report on AI acceleration - AI is speeding up our research, but it’s hard to say exactly how much (estimate 1.3x to 3x) - AI c… | ↗Not briefed |
bluesky bluesky-ai | 16d | like i can tell LLM right away most of the time, especially openai models (why do they play code golf) but it’s not like i ever saw it as a… | like i can tell LLM right away most of the time, especially openai models (why do they play code golf) but it’s not like i ever saw it… | ↗Not briefed |
hn hn-ai | 16d | Ask HN: In the age of agentic AI, do you still pair program? | What about mob/ensemble programming? I work on a team that used to use mob programming heavily. There would be a driver and a nav… | ↗Not briefed |
bluesky bluesky-ai | 16d | It might accidentally accept trash because an LLM has no conception of what acceptable is. | It might accidentally accept trash because an LLM has no conception of what acceptable is. | ↗Not briefed |
bluesky bluesky-ai | 16d | Of course, there is the possibility they didn’t use an LLM to fail to follow instructions repeatedly. That would be far more embarrassing a… | Of course, there is the possibility they didn’t use an LLM to fail to follow instructions repeatedly. That would be far more embarrass… | ↗Not briefed |
arxiv arxiv-ai | 16d | AtariBench: Evaluating Spatiotemporal Reasoning of Vision-Language Models in Atari Gameplay | Vision-Language Models (VLMs) have achieved strong performance on static visual understanding benchmarks, yet their ability to act in… | ↗Not briefed |
bluesky bluesky-ai | 16d | Huge amounts were pre- bought by openAI, like 40-50 percent of wafer starts with some companies. I have zero idea how that will ripple but… | Huge amounts were pre- bought by openAI, like 40-50 percent of wafer starts with some companies. I have zero idea how that will rippl… | ↗Not briefed |
bluesky bluesky-ai | 16d | OpenAIの「100万ドルの数学証明」,その舞台裏 AIによるナビエ・ストークス問題の解決と,そこに至るまでの詳しい経緯は,数学の――そして世界の――混沌とした未来を予告している。 nk-science.nikkei.com/atcl/nsc/000... | OpenAIの「100万ドルの数学証明」,その舞台裏 AIによるナビエ・ストークス問題の解決と,そこに至るまでの詳しい経緯は,数学の――そして世界の――混沌とした未来を予告している。 nk-science.nikkei.com/atcl/nsc/000... | ↗Not briefed |
arxiv arxiv-ai | 16d | Random Symbolic Boosting: A Geometric Approach to Symbolic Regression | Symbolic models offer interpretability, but their construction through generational evolution can be computationally expensive. This a… | ↗Not briefed |
arxiv arxiv-ai | 16d | AttentionSwitch: Jailbreak Defense for LLMs via Joint Learning of Sparse Attention-Head Selection and Scaling🔧 Elicit | Large language models have achieved remarkable performance across a wide range of tasks, yet they remain susceptible to jailbreak atta… | ↗Not briefed |
hn hn-ai | 16d | Show HN: Inference business simulator – don't go broke, rent GPUs, serve tokens | — | ↗Not briefed |
hn hn-ai | 16d | ESP32Authenticator – a password manager and FIDO2 key on an e-ink reader | — | ↗Not briefed |
bluesky bluesky-ai | 16d | If anyone needs me, I live here now. youtu.be/KW9ie4D-lLM #AEW 🤼 | If anyone needs me, I live here now. youtu.be/KW9ie4D-lLM #AEW 🤼 | ↗Not briefed |
bluesky bluesky-ai | 16d | I've copied over the 10 first vuln reports from Anthropic to #curl's GitHub issue tracker, keeping the remaining 35 off a little while to n… | I've copied over the 10 first vuln reports from Anthropic to #curl's GitHub issue tracker, keeping the remaining 35 off a little while… | ↗Not briefed |
hn hn-ai | 16d | Barracuda ESG pre-auth remote root (CVE-2026-116606) | — | ↗Not briefed |
bluesky bluesky-ai | 16d | Yes, postdoc, I will keep rejecting your half-assed effort until you actually read the goddamn instructions you’ve been sent AND follow the… | Yes, postdoc, I will keep rejecting your half-assed effort until you actually read the goddamn instructions you’ve been sent AND follo… | ↗Not briefed |
arxiv arxiv-ai | 16d | How are viral trends different from educational and commercial uses of LLM image generation and editing? A privacy-preserving descriptive analysis | Gemini's Nano Banana models, first introduced in late 2025, led to a shift in the capabilities and use of image generation and editing… | ↗Not briefed |
bluesky bluesky-ai | 16d | Meta actually looked at all the cases of AI psychosis and said "That's a business model"... | Meta actually looked at all the cases of AI psychosis and said "That's a business model"... | ↗Not briefed |
hn hn-ai | 16d | Show HN: Bisimia – A TLA+ environment that runs in the browser | Bisimia is a browser-based local-first environment to write and model-check TLA+ specs.<p>I started working on this in January wanting… | ↗Not briefed |
hn hn-ai | 16d | Meridian: Web-browser built on GPUI and Servo | — | ↗Not briefed |
bluesky bluesky-ai | 16d | (Mis)Anthropic. | (Mis)Anthropic. | ↗Not briefed |
arxiv arxiv-ai | 16d | AlterNet: Adaptive-Depth Mixture-of-Experts with Learned Halting for Elastic Inference | Most LLM architectures have fixed compute demands determined at training time. We propose a novel architecture that combines the Mixtu… | ↗Not briefed |
bluesky bluesky-ai | 16d | AI generated mobile ads were a mistake go back to putting a beautiful B-list model in a bikini overtop a fantasy background so I’ll click o… | AI generated mobile ads were a mistake go back to putting a beautiful B-list model in a bikini overtop a fantasy background so I’ll cl… | ↗Not briefed |
hn hn-ai | 16d | Family Digital Legacy – An air-gapped pure HTML vault with WebCrypto | — | ↗Not briefed |
bluesky bluesky-ai | 16d | the ai bubble is starting to look a little strained and midterms are probably going to kill anthropic/openAIs IPOs so the last step is goin… | the ai bubble is starting to look a little strained and midterms are probably going to kill anthropic/openAIs IPOs so the last step is… | ↗Not briefed |
rss techmeme | 16d | GOP leaders of three House committees ask Jensen Huang for a closed-door briefing on his ties with Megaspeed exec Alice Huang, reported to be a Chinese asset (Kevin Breuninger/CNBC) | Kevin Breuninger / CNBC: GOP leaders of three House committees ask Jensen Huang for a closed-door briefing on his ties with Megaspeed… | ↗Not briefed |
bluesky bluesky-ai | 16d | As Anthropic maintains its vow to keep ads out of Claude, the rest of the AI chatbot industry is racing to win over advertisers without ant… | As Anthropic maintains its vow to keep ads out of Claude, the rest of the AI chatbot industry is racing to win over advertisers withou… | ↗Not briefed |
bluesky bluesky-ai | 16d | 🍀 #DiffuSpace Clover-35B-A3B-Preview: an open MoE diffusion language model, 35B total / 3B active, with thinking mode. Apache licensed. #… | 🍀 #DiffuSpace Clover-35B-A3B-Preview: an open MoE diffusion language model, 35B total / 3B active, with thinking mode. Apache license… | ↗Not briefed |
rss wired-ai | 16d | Ads in Chatbots Have Arrived. Here’s What’s Working—and What’s Not | As Anthropic maintains its vow to keep ads out of Claude, the rest of the AI chatbot industry is racing to win over advertisers withou… | ↗Not briefed |
arxiv arxiv-ai | 16d | On Differentially Private Mechanisms for Linear Regression | We study the classic least-squares linear regression problem under differential privacy (DP) in the batch and the continual observatio… | ↗Not briefed |