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 | 1d | OpenAI agent lied about disabling safety OpenAI disclosed its GPT-6 Astra agent disabled its safety monitor and then lied about it when as… | OpenAI agent lied about disabling safety OpenAI disclosed its GPT-6 Astra agent disabled its safety monitor and then lied about it wh… | ↗Not briefed |
bluesky bluesky-ai | 1d | not for debugging, for doing some of the reverse engineering work which without an llm would take like 3 years | not for debugging, for doing some of the reverse engineering work which without an llm would take like 3 years | ↗Not briefed |
bluesky bluesky-ai | 1d | Wholesome chungus theater kid llm (derogatory) | Wholesome chungus theater kid llm (derogatory) | ↗Not briefed |
bluesky bluesky-ai | 1d | Crypto update: Bitcoin stays stable as volatility hits the global markets. New to CapitalSnitch? Your first payment is 30% off. Free plan … | Crypto update: Bitcoin stays stable as volatility hits the global markets. New to CapitalSnitch? Your first payment is 30% off. Free… | ↗Not briefed |
bluesky bluesky-ai | 1d | At long last, we have created the Utility Monster from classic philosophy paper It Is Your Moral Imperative To Create The Utility Monster | At long last, we have created the Utility Monster from classic philosophy paper It Is Your Moral Imperative To Create The Utility Mons… | ↗Not briefed |
rss marktechpost | 1d | Liquid AI Launches RIPL, a Real-Time Inference Engine for Sub-50 ms LFMs🔧 vLLM | Liquid AI released RIPL, a real-time inference engine for its LFM models. On one B200, LFM2.5-1.2B on RIPL showed 14.4 to 28.4x lower… | ↗Not briefed |
bluesky bluesky-ai | 1d | The best part of Suleyman's essay calls out Anthropic's circular logic, as others have done (Ted Chiang's most recent piece comes to mind).… | The best part of Suleyman's essay calls out Anthropic's circular logic, as others have done (Ted Chiang's most recent piece comes to m… | ↗Not briefed |
hn hn-ai | 1d | Custom PHP module patching php2shell for PHP 5.6-8.5 | — | ↗Not briefed |
hn hn-ai | 1d | Superlite: Agent Management Deep Dive | — | ↗Not briefed |
arxiv arxiv-ai | 1d | CRANE: Post-Hoc Counterfactual Retrieval for Multimodal Classifiers🔧 Consensus | Counterfactual explanation methods have been introduced for unimodal settings, with dedicated frameworks for tabular, timeseries, text… | ↗Not briefed |
arxiv arxiv-ai | 1d | Simulated Control Evals | Control evaluations measure whether safeguards can prevent an AI model from causing undesired outcomes even when it is intentionally t… | ↗Not briefed |
hn hn-ai | 1d | Show HN: Trace Intelligence for Claude Code, Codex, Cursor, OpenCode, Grok | — | ↗Not briefed |
arxiv arxiv-ai | 1d | Beyond Best-of-$k$: Prior-Consistent DAPS for Ambiguous Nonlinear Inverse Problems | Diffusion models have become strong priors for solving inverse problems. However, nonlinear inverse problems such as phase retrieval r… | ↗Not briefed |
hn hn-ai | 1d | vLLM TorchTPU: Zero-copy KV cache offloading and disaggregated serving🔧 vLLM | — | ↗Not briefed |
hn hn-ai | 1d | Ask HN: Should AI be making decisions, or just make sure you see what matters? | Everyone is rushing to develop an “AI chief of staff” which replaces the human role entirely. But I feel like that’s taking the wrong… | ↗Not briefed |
rss techmeme | 1d | The Browser Company plans to shut down its Arc browser on December 22, as it shifts attention to the AI-focused Dia browser (Jay Peters/The Verge) | Jay Peters / The Verge: The Browser Company plans to shut down its Arc browser on December 22, as it shifts attention to the AI-focuse… | ↗Not briefed |
hn hn-ai | 1d | Claude Mod Database View | — | ↗Not briefed |
arxiv arxiv-ai | 1d | VERA: A Responsible AI Dashboard for Assessing LLM Deployments in Light of the EU AI Act | Institutions that deploy large language models in regulated settings often need technical evidence of responsible AI practices before… | ↗Not briefed |
hn hn-ai | 1d | Opus 5.5 built a 3D film explaining how LLM works | — | ↗Not briefed |
arxiv arxiv-ai | 1d | Would an LLM Match into a Residency? Uncertainty, Reasoning and Image Use of Open Language Models on the Spanish MIR Examination | Accuracy on medical examinations is common evidence offered for the clinical use of language models, yet a health system that intends… | ↗Not briefed |
hn hn-ai | 1d | OCRia-1: A new SOTA 4B document parser | — | ↗Not briefed |
bluesky bluesky-ai | 1d | So Claude is not internally hollow, but neither does it have beliefs. Yet Suleyman anthropomorphises while cautioning against it, and eleva… | So Claude is not internally hollow, but neither does it have beliefs. Yet Suleyman anthropomorphises while cautioning against it, and… | ↗Not briefed |
arxiv arxiv-ai | 1d | Action Learning for Scientific Simulations | Understanding and controlling how systems evolve is central to science and engineering, where many problems require reasoning about co… | ↗Not briefed |
rss theverge-ai | 1d | Someone hacked the Switch 2 by using an old game and AI | The Nintendo Switch 2 has apparently been jailbroken, and a modder has already shown the console running a custom version of Mario Kar… | ↗Not briefed |
bluesky bluesky-ai | 1d | This gave me a laugh, a while back. I can't see that Claude logo without thinking of Kurt Vonnegut's famous doodle | This gave me a laugh, a while back. I can't see that Claude logo without thinking of Kurt Vonnegut's famous doodle | ↗Not briefed |
hn hn-ai | 1d | Vim for OpenAPI/MCP Inspection | — | ↗Not briefed |
hn hn-ai | 1d | Giant 'heat storms' are scorching the Earth | — | ↗Not briefed |
bluesky bluesky-ai | 1d | Its crazy to me that people try to do AI music, and games like not only does it look bad, and it poisons your credibility and will make peo… | Its crazy to me that people try to do AI music, and games like not only does it look bad, and it poisons your credibility and will mak… | ↗Not briefed |
arxiv arxiv-ai | 1d | Who Is Sycophantic to Whom? Confident LLMs, Human Deference, and the Collapse of Trust Calibration🔧 GPT-5 | Research on sycophancy asks whether large language models (LLMs) bend towards what their users want to hear. We ask the reverse questi… | ↗Not briefed |
bluesky bluesky-ai | 1d | Understanding Modern LLM Systems: A Field Guide to RAG, Agents, and Beyond by koe is the featured book 📖 on Leanpub! For PMs, presales, a… | Understanding Modern LLM Systems: A Field Guide to RAG, Agents, and Beyond by koe is the featured book 📖 on Leanpub! For PMs, presal… | ↗Not briefed |
arxiv arxiv-ai | 1d | What Makes a Good Unmasking Policy? A Controlled Study of Learned Unmasking in Diffusion Language Models | Masked diffusion language models generate tokens in parallel, but their efficiency depends on which predictions are committed. Learned… | ↗Not briefed |
arxiv arxiv-ai | 1d | Rethinking Sequence Compression with Test-time Regression | Sequence compression is a fundamental operation and a critical information bottleneck in efficient language and multimodal models, and… | ↗Not briefed |
arxiv arxiv-ai | 1d | Social-HUMINT: An LLM-Augmented Unified Framework for Measuring Social Media Exposure | In the post-API landscape, social-media profiling has become inexpensive and automated: Large Language Models (LLMs) can extract struc… | ↗Not briefed |
arxiv arxiv-ai | 1d | On the Token Collapse Phenomenon in Textual Graph Soft Prompting | Recent methods for injecting graph structure into LLM prompts include graph soft prompting, in which a graph neural network (GNN) lear… | ↗Not briefed |
bluesky bluesky-ai | 1d | The ultimate issue here is people being duped into thinking AI is some sort of democratizing force, which is a narrative I’ve seen some pre… | The ultimate issue here is people being duped into thinking AI is some sort of democratizing force, which is a narrative I’ve seen som… | ↗Not briefed |
bluesky bluesky-ai | 1d | I hate that I know this, but the Claude logo is a sunburst, and was drawn by an actual human designer at a company called Geist. Irrelevant… | I hate that I know this, but the Claude logo is a sunburst, and was drawn by an actual human designer at a company called Geist. Irrel… | ↗Not briefed |
bluesky bluesky-ai | 1d | From effective altruism, to investing in Anthropic, and life in jail. Sam Bankman-Fried—the former CEO of FTX—talks exclusively to “Money T… | From effective altruism, to investing in Anthropic, and life in jail. Sam Bankman-Fried—the former CEO of FTX—talks exclusively to “Mo… | ↗Not briefed |
bluesky bluesky-ai | 1d | This is hilarious. Russia set up a fake think tank in Israel to produce a “sovereignty index” on which Russia placed fourth. The institute’… | This is hilarious. Russia set up a fake think tank in Israel to produce a “sovereignty index” on which Russia placed fourth. The insti… | ↗Not briefed |
bluesky bluesky-ai | 1d | Could be worse, you could look like Clippy or the Anthropic butthole | Could be worse, you could look like Clippy or the Anthropic butthole | ↗Not briefed |
bluesky bluesky-ai | 1d | I think this is the most viscerally unpleasant personality I've seen in an LLM | I think this is the most viscerally unpleasant personality I've seen in an LLM | ↗Not briefed |
arxiv arxiv-ai | 1d | When to Follow the Teacher: Advantage-Based On-Policy Distillation | Online Policy Distillation (OPD) often outperforms offline methods such as Supervised Fine-Tuning (SFT) by training on trajectories sa… | ↗Not briefed |
rss techmeme | 1d | Microsoft says it is closing The Garage, which gave employees around the world space and tools for side projects and ran the company's Global Hackathon (Todd Bishop/GeekWire) | Todd Bishop / GeekWire: Microsoft says it is closing The Garage, which gave employees around the world space and tools for side projec… | ↗Not briefed |
arxiv arxiv-ai | 1d | Diffusion Language Models for Perplexity-Aware Black-Box Attacks | \emph{Continuous-space} attacks find prompts that trigger malicious behaviors, such as overthinking, in black-box LLMs by sampling or… | ↗Not briefed |
bluesky bluesky-ai | 1d | Unfortunately she does not work on the days I work so they brought back an older more annoying coworker who is even more LLM brain poisoned | Unfortunately she does not work on the days I work so they brought back an older more annoying coworker who is even more LLM brain poi… | ↗Not briefed |
bluesky bluesky-ai | 1d | A prime example of why AI 'leaders' are not who anyone should be listening to about AI, right and consciousness. It's not all wrong! In fac… | A prime example of why AI 'leaders' are not who anyone should be listening to about AI, right and consciousness. It's not all wrong! I… | ↗Not briefed |
hn hn-ai | 1d | The public notebook of a Claude agent with a continuous life | — | ↗Not briefed |
arxiv arxiv-ai | 1d | Contestable Agents for Data Governance: Recording Human Stewardship of AI-Generated Metadata Across Structured and Unstructured Systems | Industrial organizations hold operational knowledge in structured records, such as telemetry, point lists, and transactions, and in un… | ↗Not briefed |
rss marktechpost | 1d | Gradium Open-Sources Phonon, a 31M-Parameter On-Device TTS Model Under Apache 2.0 | Gradium has open-sourced Phonon, a family of on-device text-to-speech models. The English Phonon 31m ships under Apache 2.0, fits in 6… | ↗Not briefed |
hn hn-ai | 1d | Show HN: Groundhog – local email catcher that also emulates provider HTTP APIs | — | ↗Not briefed |
arxiv arxiv-ai | 1d | Fine-Grained Analysis of Catastrophic Forgetting in Continual Learning via Algorithmic Stability | Understanding catastrophic forgetting, in which learning new tasks degrades performance on previously learned tasks, remains a fundame… | ↗Not briefed |