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 |
|---|---|---|---|---|
hn hn-ai | 1mo | Askholes and Low Effort AI Answers | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | I sort of think the plan here is to get it nationalized before the bubble bursts. | I sort of think the plan here is to get it nationalized before the bubble bursts. | ↗Not briefed |
arxiv arxiv-ai | 1mo | Entropy Regularization: A Free Correction to Cross-Entropy for Verified Demonstrations | Large language models are often post-trained on expert demonstrations using cross-entropy (CE), even when the downstream objective is… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Another one: why would we pay for extra capacity? Our server sits around 33% of capacity for the vast majority of the time. *Normal* LLM re… | Another one: why would we pay for extra capacity? Our server sits around 33% of capacity for the vast majority of the time. *Normal* L… | ↗Not briefed |
arxiv arxiv-ai | 1mo | HARDEN: Constrained Evolutionary Search for Harder, Answer-Preserving Evaluation Cases | Language models are often evaluated on curated benchmarks that underrepresent the complexity of enterprise deployments. We introduce H… | ↗Not briefed |
bluesky bluesky-ai | 1mo | waterfox implementing a built-in setting that disables LLM search results regardless of which search engine youre using AND TURNING IT ON B… | waterfox implementing a built-in setting that disables LLM search results regardless of which search engine youre using AND TURNING IT… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Koi.rest – watch some fish and regain your balance | Hi there. My name is Paul. I’m a developer, an explorer and I’ve got ADHD. As of the beginning of August, I’m also unemployed.<p>This… | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI did not "go rogue". Despite what sci-fi movies tell you, AI is an algorithm. A human creation. It does not possess will of its own. … | OpenAI did not "go rogue". Despite what sci-fi movies tell you, AI is an algorithm. A human creation. It does not possess will of its… | ↗Not briefed |
bluesky bluesky-ai | 1mo | sharing with screenshot of quoted post for posterity -- kind of amusing overlap here with the trump admin's crackdown on the word "bias" in… | sharing with screenshot of quoted post for posterity -- kind of amusing overlap here with the trump admin's crackdown on the word "bia… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Atelier: Learning Local Self-Supervised Features for CryoEM Volumes via Hypernetworks | CryoEM map interpretation requires features that are spatially localized, consistent across samples, and informative across spatial sc… | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI breach of Australian government linked to wider AI hacking campaign ft.trib.al/68nsZH5 | OpenAI breach of Australian government linked to wider AI hacking campaign ft.trib.al/68nsZH5 | ↗Not briefed |
hn hn-ai | 1mo | Meta's Muse page can't even pass basic privacy requirements | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Every recent “model safety” incident was the result of careless (and potentially malicious) cybersecurity controls, not “out of control” AI… | Every recent “model safety” incident was the result of careless (and potentially malicious) cybersecurity controls, not “out of contro… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Just a few days in, my mom escaped containment from the AL facility faster than an unaligned AI model. She is back there with additional gu…🔧 Hugging Face | Just a few days in, my mom escaped containment from the AL facility faster than an unaligned AI model. She is back there with addition… | ↗Not briefed |
bluesky bluesky-ai | 1mo | 2-4 are jagon-y and difficult to parse without knowing the lingo, but #1 seems pretty bang on to me. It's an oversimplification of what an … | 2-4 are jagon-y and difficult to parse without knowing the lingo, but #1 seems pretty bang on to me. It's an oversimplification of wha… | ↗Not briefed |
bluesky bluesky-ai | 1mo | A not-insignificant percentage of people who hate AI are gleefully assisting companies like Anthropic, OpenAI and xAI by parroting their in… | A not-insignificant percentage of people who hate AI are gleefully assisting companies like Anthropic, OpenAI and xAI by parroting the… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Kinda fun how the way these systems “go rogue” always benefit openAI They never just decide everyone gets unlimited tokens forever or sta… | Kinda fun how the way these systems “go rogue” always benefit openAI They never just decide everyone gets unlimited tokens forever o… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Public Browser MCP – +48% speed -33% token/session vs. AgentBrowser | (a repost due to gen-ai block | sry guys English is not my native language)<p>me: solo dev from Hamburg (ahoi) started months ago - co… | ↗Not briefed |
bluesky bluesky-ai | 1mo | OpenAI and the rest need to be sued out of existence. This is on top of the story of one of the OpenAI agents hacking into an Australian Me… | OpenAI and the rest need to be sued out of existence. This is on top of the story of one of the OpenAI agents hacking into an Australi… | ↗Not briefed |
rss techmeme | 1mo | Anthropic commits to spending $11.6B over seven years on Akamai cloud services and secures an option to take a stake of up to 5%; AKAM jumps 17%+ after hours (Harshita Mary Varghese/Reuters) | Harshita Mary Varghese / Reuters: Anthropic commits to spending $11.6B over seven years on Akamai cloud services and secures an option… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I think it's very funny that OpenAI has overcorrected on "AI generators can't make coherent text" by making every single image their genera… | I think it's very funny that OpenAI has overcorrected on "AI generators can't make coherent text" by making every single image their g… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Illinois is doing some things chicago.suntimes.com/politics/202... | Illinois is doing some things chicago.suntimes.com/politics/202... | ↗Not briefed |
bluesky bluesky-ai | 1mo | Is the idea that it was instructed to gather medical information, and it moved beyond the parameters and accessed private information bc th… | Is the idea that it was instructed to gather medical information, and it moved beyond the parameters and accessed private information… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Thinking Less to Simulate Better: Intuitive Prompting Improves LLM Agents Simulating Individual Social Media Reactions, Including Unfamiliar Content | Platform policies are increasingly tested on artificial users, making agent fidelity important. Yet convincing fake profiles could als… | ↗Not briefed |
bluesky bluesky-ai | 1mo | I’m curious to know how this information would benefit the OpenAI system. It hacked an Australian government database for…what reason, exac… | I’m curious to know how this information would benefit the OpenAI system. It hacked an Australian government database for…what reason,… | ↗Not briefed |
bluesky bluesky-ai | 1mo | based on what I have seen of the proposed anti-athropomorphization standards, I think the third attempt fails for saying "the LLM was instr… | based on what I have seen of the proposed anti-athropomorphization standards, I think the third attempt fails for saying "the LLM was… | ↗Not briefed |
bluesky bluesky-ai | 1mo | they will be fed into their latest ai model | they will be fed into their latest ai model | ↗Not briefed |
arxiv arxiv-ai | 1mo | Probing Stability-Plasticity Tradeoffs in Agent Memory through Cognitive Experimental Paradigms | Agent memory systems are increasingly used to maintain long-term user preferences, task states and evolving facts, but current evaluat… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Like if you replace model, ai, or llm in these articles with software they just turn into “we had a bug and it was bad” type articles which… | Like if you replace model, ai, or llm in these articles with software they just turn into “we had a bug and it was bad” type articles… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Auditing Latent-Space Monitors for Autonomous Driving | Runtime failure monitors can use a model's internal representations to anticipate failures. We audit this monitoring strategy across t… | ↗Not briefed |
hn hn-ai | 1mo | IR (2022) | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | Beginning to suspect Anthropic saw Stancil’s viewer count and tenacity at promoting a line, and hired him for this performance. | Beginning to suspect Anthropic saw Stancil’s viewer count and tenacity at promoting a line, and hired him for this performance. | ↗Not briefed |
arxiv arxiv-ai | 1mo | Dynamic Regret in Online Convex Optimization with Indicator Switching Costs | We study dynamic regret in online convex optimization with an \emph{indicator switching cost}: a fixed penalty incurred whenever two c… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Had Paul Kedrosky back on the show to talk about how AI is distorting the bond markets to the point that it’s changing the price on US Trea… | Had Paul Kedrosky back on the show to talk about how AI is distorting the bond markets to the point that it’s changing the price on US… | ↗Not briefed |
hn hn-ai | 1mo | Hail is a multibillion-dollar problem – and could get worse | — | ↗Not briefed |
bluesky bluesky-ai | 1mo | You’d think the most effective ‘guardrail’ against the deleterious affects of AI is not let so many of the bloody awful data centres be bui… | You’d think the most effective ‘guardrail’ against the deleterious affects of AI is not let so many of the bloody awful data centres b… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Proportional Representation in Temporal Voting with Ranked Preferences | We study proportional representation in temporal voting, where one candidate is selected in each round. While prior work has focused o… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Translation: "We want to make your children unable to think so when they're adults we can sell our LLM app to helpless idiots who are unabl… | Translation: "We want to make your children unable to think so when they're adults we can sell our LLM app to helpless idiots who are… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Könnte ein LLM das in gut aussehend? Sicher. Aber hier wird noch unnötig viel Zeit mit MSPaint auf zwei-Fav-Kaspereien vergeudet. Solide Ha… | Könnte ein LLM das in gut aussehend? Sicher. Aber hier wird noch unnötig viel Zeit mit MSPaint auf zwei-Fav-Kaspereien vergeudet. Soli… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Rank-Reliable Teacher-Guided Fitness Approximation for Expensive Evolutionary Optimization: A TinyML Architecture Search Study | Expensive evolutionary search does not always need an exact fitness estimate for every candidate. It often needs a reliable answer to… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Have OpenAI researchers considered wiping off the word they inscribed into the wet clay of ChatGPT’s forehead? That sometimes works. www.…🔧 ChatGPT | Have OpenAI researchers considered wiping off the word they inscribed into the wet clay of ChatGPT’s forehead? That sometimes works.… | ↗Not briefed |
hn hn-ai | 1mo | Vibe Coding Production Kit – a production workflow for AI coding agents | — | ↗Not briefed |
hn hn-ai | 1mo | Show HN: PipeFitter – Quick, Browser-Based, Parametric Pipe Connectors | Ever need to connect two pipes of slightly different sizes together? I kept opening CAD and spending more minutes tweaking designs tha… | ↗Not briefed |
bluesky bluesky-ai | 1mo | So when is the openai guy getting arrested? | So when is the openai guy getting arrested? | ↗Not briefed |
bluesky bluesky-ai | 1mo | This is obviously true but I think we mean a different thing when we say an LLM is biased---we don't mean E(\hat{\theta} - \theta) != 0, we… | This is obviously true but I think we mean a different thing when we say an LLM is biased---we don't mean E(\hat{\theta} - \theta) !=… | ↗Not briefed |
bluesky bluesky-ai | 1mo | AI security is turning into AI versus AI. A whitehat firm used Claude to attack OpenAI. Security Now looks at what that means for testing… | AI security is turning into AI versus AI. A whitehat firm used Claude to attack OpenAI. Security Now looks at what that means for te… | ↗Not briefed |
rss techmeme | 1mo | At a White House summit, Xi Jinping says the US and China have "the capability and responsibility to develop and manage AI for good" as "leading nations" in AI (Mallory Wilson/The Hill) | Mallory Wilson / The Hill: At a White House summit, Xi Jinping says the US and China have “the capability and responsibility to develo… | ↗Not briefed |
hn hn-ai | 1mo | Show HN: Canary (YC) – Independent verification for AI code | Hey HN, we are Aakash and Viswesh and we are building Canary (<a href="https://www.runcanary.ai/">https://www… | ↗Not briefed |
arxiv arxiv-ai | 1mo | Benchy: towards a universal language for task-oriented AI benchmarks | Benchy is a semantic language and execution engine for benchmarking AI programs. A benchmark is completely specified by a program, a s… | ↗Not briefed |
bluesky bluesky-ai | 1mo | Burn OpenAI to the ground. | Burn OpenAI to the ground. | ↗Not briefed |