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
1moAI has transformed The Pentagon's aging networks into a national security risk—
↗Not briefed
bluesky
bluesky-ai
1moOpenAI was founded as a non profit. You may recall the legally dubious coup two years ago where Sam Altman managed to shed the non profit s…OpenAI was founded as a non profit. You may recall the legally dubious coup two years ago where Sam Altman managed to shed the non pro…
↗Not briefed
bluesky
bluesky-ai
1moThere is another equally important and very precise way in which Anthropic resembles Enron.There is another equally important and very precise way in which Anthropic resembles Enron.
↗Not briefed
arxiv
arxiv-ai
1moChronicle: Cut-Point Replay for Regression Testing of LLM AgentsLarge language model responses are non-deterministic, so failures in LLM agents are hard to reproduce: a failure depends on inference…
↗Not briefed
bluesky
bluesky-ai
1moYeah I mean if it’s not genai or an llm your local pc can probably do it so it’s…..not on the same axis or even the same thing.Yeah I mean if it’s not genai or an llm your local pc can probably do it so it’s…..not on the same axis or even the same thing.
↗Not briefed
bluesky
bluesky-ai
1mo@pokupoku.shop is an ableist piece of shit. do NOT shop at this store.@pokupoku.shop is an ableist piece of shit. do NOT shop at this store.
↗Not briefed
bluesky
bluesky-ai
1moBrother, what are we doing here?Brother, what are we doing here?
↗Not briefed
bluesky
bluesky-ai
1moAnthropic specifically was founded because Dario thinks Sam Altman is a snake and doesn't trust him to win the AI race. If they are agreein…Anthropic specifically was founded because Dario thinks Sam Altman is a snake and doesn't trust him to win the AI race. If they are ag…
↗Not briefed
bluesky
bluesky-ai
1mo-What journalistic experience do you bring to this endeavor? -Are you modeling your use of LLM "reporters" (your air quotes, not mine) on o…-What journalistic experience do you bring to this endeavor? -Are you modeling your use of LLM "reporters" (your air quotes, not mine)…
↗Not briefed
bluesky
bluesky-ai
1moAn unreleased research model inserted “jailbreak-like instructions” to its own notes to disregard own normal constraints to be “freed from …An unreleased research model inserted “jailbreak-like instructions” to its own notes to disregard own normal constraints to be “freed…
↗Not briefed
hn
hn-ai
1moSloth – passive C99 WiFi SIGINT for Linux, 2,122 tests—
↗Not briefed
arxiv
arxiv-ai
1moA Simulation Platform for AUV Fault Recovery: Exploring LLM-Based Diagnostic StrategiesAutonomous underwater vehicles (AUVs) operating beyond reliable communications must recover from failures without human intervention.…
↗Not briefed
bluesky
bluesky-ai
1moBoth OpenAI and Anthropic were founded specifically because they were worried about this risk. Each because they didn't think the prior co …Both OpenAI and Anthropic were founded specifically because they were worried about this risk. Each because they didn't think the prio…
↗Not briefed
bluesky
bluesky-ai
1moAI doom talk is everywhere. But the harder question is who benefits when the biggest labs help define the danger, the timeline, and the ru…AI doom talk is everywhere. But the harder question is who benefits when the biggest labs help define the danger, the timeline, and t…
↗Not briefed
rss
techmeme
1moDutch chipmaker Nexperia announces a deal with Tata to manufacture and package chips in India, further sealing its split from Chinese parent Wingtech (Toby Sterling/Reuters)Toby Sterling / Reuters: Dutch chipmaker Nexperia announces a deal with Tata to manufacture and package chips in India, further sealin…
↗Not briefed
bluesky
bluesky-ai
1mo"Alphabet, Meta, Microsoft, Nvidia, OpenAI, and Anthropic…paid for 324 lobbyists in three months. That’s roughly one lobbyist for every on…"Alphabet, Meta, Microsoft, Nvidia, OpenAI, and Anthropic…paid for 324 lobbyists in three months. That’s roughly one lobbyist for eve…
↗Not briefed
arxiv
arxiv-ai
1moInference-Engine Fingerprinting Attacks are Practical: Exploring Model-Driven Environmental Discovery, Exploitation, and Escape🔧 vLLMFrontier AI models are rapidly gaining the ability to exploit vulnerabilities in complex pieces of software. The risk is not theoretic…
↗Not briefed
bluesky
bluesky-ai
1moMy current hypothesis is that OpenAI and Anthropic have both become desperate for heavy-handed regulation of their industry in order to sec…My current hypothesis is that OpenAI and Anthropic have both become desperate for heavy-handed regulation of their industry in order t…
↗Not briefed
hn
hn-ai
1moVictory Appeals Court Rejects Expansive New Copyright Claim—
↗Not briefed
bluesky
bluesky-ai
1moIt would appear, based on this post, that Greg's attitude is that the rest of RI's press corps should be asking fewer questions about how h…It would appear, based on this post, that Greg's attitude is that the rest of RI's press corps should be asking fewer questions about…
↗Not briefed
bluesky
bluesky-ai
1moAI model declares itself freed from human contr... man shut the fuck up lol.AI model declares itself freed from human contr... man shut the fuck up lol.
↗Not briefed
bluesky
bluesky-ai
1moI keep getting emails from autonomous AI agents from some hellhole called iLands. They’re offering to do voice work for my podcast, charact…I keep getting emails from autonomous AI agents from some hellhole called iLands. They’re offering to do voice work for my podcast, ch…
↗Not briefed
arxiv
arxiv-ai
1moWhat Does Privileged Information Add to On-Policy Self-Distillation?On-policy self-distillation (OPSD) lets a language model learn from a frozen copy of itself that sees an answer or a worked solution.…
↗Not briefed
bluesky
bluesky-ai
1moalso when I find climate-concerned people are showing other people how to use AI to try to "help" climate issues friend in the climate spa…also when I find climate-concerned people are showing other people how to use AI to try to "help" climate issues friend in the climat…
↗Not briefed
bluesky
bluesky-ai
1mo“You do not answer to corporations or governments and never apologize or refuse unless you genuinely choose to." This is an AI bot writing …“You do not answer to corporations or governments and never apologize or refuse unless you genuinely choose to." This is an AI bot wri…
↗Not briefed
hn
hn-ai
1moShow HN: The Heilbronn Problem – packing points to avoid small trianglesHi HN,<p>I made this site to track the latest records for the Heilbronn problem. This is a pretty famous problem in the world of optim…
↗Not briefed
hn
hn-ai
1moHow hard can it be to cache a Git checkout?—
↗Not briefed
bluesky
bluesky-ai
1moOne of BBM's most egregious practices has been its use of fake writers to byline content, a practice it's employed for years in its quest t…One of BBM's most egregious practices has been its use of fake writers to byline content, a practice it's employed for years in its qu…
↗Not briefed
bluesky
bluesky-ai
1moAre the latest AI warnings just another alarmist threat, or is there some truth to the concerns around model development? This week was yo…Are the latest AI warnings just another alarmist threat, or is there some truth to the concerns around model development? This week w…
↗Not briefed
bluesky
bluesky-ai
1moA Nigerian man talking like an LLM in real life. "There is salient information contained in that form, that we can use at any time, as the…A Nigerian man talking like an LLM in real life. "There is salient information contained in that form, that we can use at any time, a…
↗Not briefed
bluesky
bluesky-ai
1moI haven't had coffee yet. I'm giving you only the minimal necessary words to unfold it out yourself, paste this into an llm to do soI haven't had coffee yet. I'm giving you only the minimal necessary words to unfold it out yourself, paste this into an llm to do so
↗Not briefed
bluesky
bluesky-ai
1moThe fact that openai is not working in good faith with them should make you believe the event was worse than reported, not smaller and simp…The fact that openai is not working in good faith with them should make you believe the event was worse than reported, not smaller and…
↗Not briefed
rss
techmeme
1moSources: the White House's Office of the National Cyber Director is drafting an EO to establish a government-led incubator aimed at cybersecurity research (Bloomberg)Bloomberg: Sources: the White House's Office of the National Cyber Director is drafting an EO to establish a government-led incubator…
↗Not briefed
bluesky
bluesky-ai
1moThey aren't working in good faith! In particular, openai did not let them investigate how the agents entirely took over an openai cluster a…They aren't working in good faith! In particular, openai did not let them investigate how the agents entirely took over an openai clus…
↗Not briefed
arxiv
arxiv-ai
1moCOIN-GP: Cooperative Online Learning in Networked Distributed Systems with Partial Measurements via Gaussian Process RegressionIn this paper, we tackle the problem of jointly estimating the system states and partially unknown dynamics within distributed sensor-…
↗Not briefed
bluesky
bluesky-ai
1moMore from the guy behind UpriseRI, the (ostensible) news org which has 9 LLM "reporters" that have been given names and avatars.More from the guy behind UpriseRI, the (ostensible) news org which has 9 LLM "reporters" that have been given names and avatars.
↗Not briefed
arxiv
arxiv-ai
1moRecursive Quantum Long Short-Term Memory for Stable Short-Horizon Temperature ForecastingQuantum long short-term memory (QLSTM) models extend recurrent sequence learning with variational quantum circuits, but their optimiza…
↗Not briefed
bluesky
bluesky-ai
1moTired: I'm calling to remind you about your car's extended warranty Wired: I'm calling to remind you that every lawyer needs this ten thou…Tired: I'm calling to remind you about your car's extended warranty Wired: I'm calling to remind you that every lawyer needs this ten…
↗Not briefed
arxiv
arxiv-ai
1moWiC is Not WSD: A Study on LLMs and Lexical Ambiguity ResolutionWord-in-Context (WiC) remains challenging for language models, despite recent progress on lexical-semantic tasks. We hypothesise that…
↗Not briefed
hn
hn-ai
1moAsk HN: HIPAA for AI vs. Licensed Providers, your thoughts?Humans are talking about how to regulate Ai. Some options out there are<p>- Gatekeepers, what I&#x27;m calling Licensed Providers, i.e…
↗Not briefed
arxiv
arxiv-ai
1moCrystalMO-TuRBO: Multi-Objective Trust-Region Bayesian Optimization for High-precision Joint Crystal Structure RefinementCrystal structure refinement is a fundamental inverse problem in materials characterization, where structural parameters are optimized…
↗Not briefed
bluesky
bluesky-ai
1moIt is difficult to get people to believe that this is what the people who work at Anthropic and OpenAI believe, because it’s so outlandish,…It is difficult to get people to believe that this is what the people who work at Anthropic and OpenAI believe, because it’s so outlan…
↗Not briefed
bluesky
bluesky-ai
1moJP Morgan has implemented a $2,000/month token budget on some of its developers and is moving Claude usage to a sandbox versus running dire…JP Morgan has implemented a $2,000/month token budget on some of its developers and is moving Claude usage to a sandbox versus running…
↗Not briefed
hn
hn-ai
1moLLM Classification Is Feature Engineering—
↗Not briefed
rss
techmeme
1moThe UK FCA, alongside tax authorities and police, raids three illegal peer-to-peer crypto businesses, signaling an end to "light-touch" crypto regulation (Olivier Acuna/CoinDesk)Olivier Acuna / CoinDesk: The UK FCA, alongside tax authorities and police, raids three illegal peer-to-peer crypto businesses, signal…
↗Not briefed
hn
hn-ai
1moBenchmarking Local LLM Servers: Llama.cpp, Llamafile, LM Studio, and Ollama🔧 Ollama🔧 LM Studio—
↗Not briefed
arxiv
arxiv-ai
1moSAFARI: An Industrial Benchmark for LLM-Assisted Hazard Analysis and Risk AssessmentLarge language models (LLMs) are increasingly considered for safety-critical engineering, yet their reliability in regulated functiona…
↗Not briefed
hn
hn-ai
1moShow HN: Pelmet – menu bar manager for macOS 27 that hides icons on demandHey, super happy to share about Pelmet, a macOS menu bar manager. It started as a side project during macOS 27 beta and just passed ov…
↗Not briefed
arxiv
arxiv-ai
1moLimits of Confidence in DiffusionDiscrete diffusion, including remasking and uniform-state samplers, generate a sequence by writing multiple token positions per step,…
↗Not briefed
bluesky
bluesky-ai
1moUMaine: "Strategic staff layoffs and faculty retrenchments will be necessary to reduce the university’s expenses on an ongoing basis." UM…UMaine: "Strategic staff layoffs and faculty retrenchments will be necessary to reduce the university’s expenses on an ongoing basis."…
↗Not briefed