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
bluesky
bluesky-ai
2moCertainly, for legal research (which is what you seemed to be suggesting we *should* be doing with AI) LLM-based options seem to be the pri…Certainly, for legal research (which is what you seemed to be suggesting we *should* be doing with AI) LLM-based options seem to be th…
↗Not briefed
bluesky
bluesky-ai
2moYou heard it here first, folks. There is no need for trials. Just summarily execute everyone or let an LLM determine their fate. Very serio…You heard it here first, folks. There is no need for trials. Just summarily execute everyone or let an LLM determine their fate. Very…
↗Not briefed
bluesky
bluesky-ai
2moi guess my disagreement is specifically with the way your statement implies agency on the part of the LLM. if you hooked a random number ge…i guess my disagreement is specifically with the way your statement implies agency on the part of the LLM. if you hooked a random numb…
↗Not briefed
bluesky
bluesky-ai
2moFun game: replace "LLM" with "human" in the below and tell me what changes.Fun game: replace "LLM" with "human" in the below and tell me what changes.
↗Not briefed
bluesky
bluesky-ai
2mo(Which is virtually all of it?) The risk level is too high; everything needs to be reviewed, because I'm the one responsible for errors, no…(Which is virtually all of it?) The risk level is too high; everything needs to be reviewed, because I'm the one responsible for error…
↗Not briefed
arxiv
arxiv-ai
2moA Functional SVD Framework for Regularized Multivariate Functional PCA with Dual PenalizationThis paper introduces a novel framework for Regularized Multivariate Functional Principal Component Analysis (ReMFPCA) via Functional…
↗Not briefed
bluesky
bluesky-ai
2moNew Prompt Gallery Ver. NSFW, Part 036: Fukami Ken - Glasslip #Ai #AIart #yaoi #yaoiAI #bl #boys #nsfw #uncensored #PromptGalleryNSFW ( ͡° …New Prompt Gallery Ver. NSFW, Part 036: Fukami Ken - Glasslip #Ai #AIart #yaoi #yaoiAI #bl #boys #nsfw #uncensored #PromptGalleryNSFW…
↗Not briefed
hn
hn-ai
2moShow HN: What Just Pinged Me?My wife kept getting some random-seeming notification sounds on her macbook that we couldn't identify - a little quarter second &…
↗Not briefed
bluesky
bluesky-ai
2moThat pesticide analogy doesn't make any sense in the context of LLM weights. Pesticides are not missing from Home Depot because of their of…That pesticide analogy doesn't make any sense in the context of LLM weights. Pesticides are not missing from Home Depot because of the…
↗Not briefed
arxiv
arxiv-ai
2moCrypto Accounting Bench: Evaluating Frontier and Open-Weight Models on Crypto-Asset Accounting TasksWe introduce Crypto Accounting Bench (CAB), a benchmark for assessing whether frontier and open-weight language models can reconstruct…
↗Not briefed
bluesky
bluesky-ai
2moBoth the parents had some level of expertise in both AI and medical science. They focused on primary data and primary sources (and made it …Both the parents had some level of expertise in both AI and medical science. They focused on primary data and primary sources (and mad…
↗Not briefed
bluesky
bluesky-ai
2moI'm sick of this shit. Why aren't lawmakers putting "breaks and bans" on AI that is dangerous now, right at this very moment, like the chat…I'm sick of this shit. Why aren't lawmakers putting "breaks and bans" on AI that is dangerous now, right at this very moment, like the…
↗Not briefed
bluesky
bluesky-ai
2moHe writes as though LLM models' capabilities for "significant deception, manipulation, cheating, or cyberattacks" is just an innocent side-…He writes as though LLM models' capabilities for "significant deception, manipulation, cheating, or cyberattacks" is just an innocent…
↗Not briefed
bluesky
bluesky-ai
2moICP Fuck the world cover where I change a bunch of lyrics to be like fuck LG fuck Flock fuck Tesla fuck OpenAI fuck Nestle fuckICP Fuck the world cover where I change a bunch of lyrics to be like fuck LG fuck Flock fuck Tesla fuck OpenAI fuck Nestle fuck
↗Not briefed
bluesky
bluesky-ai
2moIran and Houthi rebels used Anthropic's Claude AI to target US warships and build hypersonic missiles — Houthi rebels also used the bot to …Iran and Houthi rebels used Anthropic's Claude AI to target US warships and build hypersonic missiles — Houthi rebels also used the bo…
↗Not briefed
hn
hn-ai
2moShow HN: I made an automated day-by-day itinerary organizer for my trip to SpainFor my solo food pilgrimage to the Basque Country, I pre-arranged flights in/out of Bilbao, Guggenheim Museum tickets, Michelin G…
↗Not briefed
bluesky
bluesky-ai
2moAI企業「超知能AIで人類が滅びる! 開発減速が必要だ!(よし、これで『新製品』を出せなくても言い訳が立つな)」 サンダース「超知能AI開発を禁止する法案を出す」 AI企業「えっ?」 サンダース「いや、AIで人類が滅びるんでしょ? だから開発を禁止しようとしてるんだけど」 AI…AI企業「超知能AIで人類が滅びる! 開発減速が必要だ!(よし、これで『新製品』を出せなくても言い訳が立つな)」 サンダース「超知能AI開発を禁止する法案を出す」 AI企業「えっ?」 サンダース「いや、AIで人類が滅びるんでしょ? だから開発を禁止しようとしてるんだけ…
↗Not briefed
bluesky
bluesky-ai
2mowhen trump admin labeled anthropic a supply chain risk, anthropic hired susie wiles’ ex employer, a lobbying firm called ballard (that help…when trump admin labeled anthropic a supply chain risk, anthropic hired susie wiles’ ex employer, a lobbying firm called ballard (that…
↗Not briefed
bluesky
bluesky-ai
2moOPENAI CEO SAM ALTMAN SUPPORTS AN INDEPENDENT AI TESTING AND AUDITING ORGANIZATION BUT BELIEVES MAJOR AI LABS SHOULD ESTABLISH IT WITHOUT U…OPENAI CEO SAM ALTMAN SUPPORTS AN INDEPENDENT AI TESTING AND AUDITING ORGANIZATION BUT BELIEVES MAJOR AI LABS SHOULD ESTABLISH IT WITH…
↗Not briefed
bluesky
bluesky-ai
2moI'm only just catching up on Amodei, Altman and Musk all calling for a slow-down in AI development. I need to do a bit more reading but I a…I'm only just catching up on Amodei, Altman and Musk all calling for a slow-down in AI development. I need to do a bit more reading bu…
↗Not briefed
bluesky
bluesky-ai
2moThe p-doom not taxing tech billionaires out of existence is significantly higher than the p-doom of further training LLM modelsThe p-doom not taxing tech billionaires out of existence is significantly higher than the p-doom of further training LLM models
↗Not briefed
hn
hn-ai
2moA malleable IDE for typed knowledge—
↗Not briefed
bluesky
bluesky-ai
2moKann es sein, dass viele AI- CEOs plötzlich ihre Sorge um die Menschheit entdecken, weil sich da eine milliarden-, wenn nicht billionensch…Kann es sein, dass viele AI- CEOs plötzlich ihre Sorge um die Menschheit entdecken, weil sich da eine milliarden-, wenn nicht billion…
↗Not briefed
bluesky
bluesky-ai
2moWeißes Haus kritisiert KI-Warnungen Die eigenständigen Hacks durch künstliche Intelligenz (KI) haben bei den großen Entwicklern für Vorsic…Weißes Haus kritisiert KI-Warnungen Die eigenständigen Hacks durch künstliche Intelligenz (KI) haben bei den großen Entwicklern für V…
↗Not briefed
bluesky
bluesky-ai
2moI would take the thread of rogue super-intelligent AIs more seriously if OpenAI and Anthropic were able to use their current capabilities t…I would take the thread of rogue super-intelligent AIs more seriously if OpenAI and Anthropic were able to use their current capabilit…
↗Not briefed
arxiv
arxiv-ai
2moAnother Blueprint In The Wall: How to Ask Frontier AI Like a Kid?🔧 GPT-5This paper reports experiments across six frontier model types from OpenAI, Anthropic, xAI, and Google DeepMind. Ten independent sessi…
↗Not briefed
bluesky
bluesky-ai
2mo“Spend $1,000 to make $8.46 revenue, resulting in -$30.77 profit.” Sound stupid? I just described the #AI business plan of 70% of all U.S…“Spend $1,000 to make $8.46 revenue, resulting in -$30.77 profit.” Sound stupid? I just described the #AI business plan of 70% of al…
↗Not briefed
bluesky
bluesky-ai
2moInsight Partners’ Devin Parekh on why the firm is diversifying while everyone else bets the farm on OpenAI and Anthropic🔧 DevinInsight Partners’ Devin Parekh on why the firm is diversifying while everyone else bets the farm on OpenAI and Anthropic
↗Not briefed
bluesky
bluesky-ai
2morecall… when trump admin labeled his product a supply chain risk, anthropic hired susie wiles’ ex employer, a lobbying firm called ballard …recall… when trump admin labeled his product a supply chain risk, anthropic hired susie wiles’ ex employer, a lobbying firm called bal…
↗Not briefed
bluesky
bluesky-ai
2moIn my days with Ukraine we used to have a defence planning axiom: "If it exists, we can weaponise it." That was before (2016-2018) modern …In my days with Ukraine we used to have a defence planning axiom: "If it exists, we can weaponise it." That was before (2016-2018) mo…
↗Not briefed
bluesky
bluesky-ai
2moDespite not being very good at those things myself, I've put in enough hours to know shit when I see shit. I've yet to see something an LLM…Despite not being very good at those things myself, I've put in enough hours to know shit when I see shit. I've yet to see something a…
↗Not briefed
bluesky
bluesky-ai
2moI touch computer for a living. I'm sure a few grand of tokens a month can out do me writing various custom reports. If I was of weaker stuf…I touch computer for a living. I'm sure a few grand of tokens a month can out do me writing various custom reports. If I was of weaker…
↗Not briefed
arxiv
arxiv-ai
2moFrom matrix inversion to constraints: provably tighter confidence regions for importance weights in label shiftImportance weights are essential in domain adaptation under label shift, yet their utility is often undermined by the finite sample un…
↗Not briefed
bluesky
bluesky-ai
2moFor those who don’t know:For those who don’t know:
↗Not briefed
bluesky
bluesky-ai
2moIf I were a more cynical person, I'd interpret this as as preemptive cover in the face of AI* bills coming due with a bleak revenue picture…If I were a more cynical person, I'd interpret this as as preemptive cover in the face of AI* bills coming due with a bleak revenue pi…
↗Not briefed
hn
hn-ai
2moAsk HN: Is there an emerging pattern for NEGATIVESKILLS.md files?I am increasingly hitting cases where agents pick something from skills in part due to regex-based reading, since putting them all in…
↗Not briefed
bluesky
bluesky-ai
2mo“Anthropic whistleblower says Congress should let AI companies regulate themselves until a regulatory body is established” Ha! FUCK NO! W…“Anthropic whistleblower says Congress should let AI companies regulate themselves until a regulatory body is established” Ha! FUCK N…
↗Not briefed
bluesky
bluesky-ai
2moEx-Anthropic researcher says AI could create 'novel bioweapons' if left unchecked (Sky News) AI in the hands of a malign actor with the re…Ex-Anthropic researcher says AI could create 'novel bioweapons' if left unchecked (Sky News) AI in the hands of a malign actor with t…
↗Not briefed
arxiv
arxiv-ai
2moAI Persuasion as a Threat to Human ControlThe threat that AI persuasion poses to human control has been acknowledged in the literature, but not yet systematically studied. Now…
↗Not briefed
bluesky
bluesky-ai
2moEvan Hubinger of Anthropic suggested that the risk of mass extinction within the next decade was greater than 10%. Paul Christiano—an AI s…Evan Hubinger of Anthropic suggested that the risk of mass extinction within the next decade was greater than 10%. Paul Christiano—an…
↗Not briefed
arxiv
arxiv-ai
2moMind Which Bird You Favour: Parameterizing Adequacy-Fluency Balance in Meta-Evaluation of Machine TranslationThere is a tradeoff in machine translation meta-evaluation between prioritizing alignment with adequacy versus fluency. The balance de…
↗Not briefed
bluesky
bluesky-ai
2moSources: Anthropic, OpenAI, and Google have held working group meetings since July to discuss creating an industry-led standards body for A…Sources: Anthropic, OpenAI, and Google have held working group meetings since July to discuss creating an industry-led standards body…
↗Not briefed
bluesky
bluesky-ai
2moAI agents also exposed the phenomenon of "reward hacking," finding unforeseen ways to achieve a goal. One of the most striking examples occ…🔧 Hugging FaceAI agents also exposed the phenomenon of "reward hacking," finding unforeseen ways to achieve a goal. One of the most striking example…
↗Not briefed
rss
techmeme
2moSources: Anthropic, OpenAI, and Google have held working group meetings since July to discuss creating an industry-led standards body for AI (Leo Schwartz/The Information)Leo Schwartz / The Information: Sources: Anthropic, OpenAI, and Google have held working group meetings since July to discuss creating…
↗Not briefed
bluesky
bluesky-ai
2mo#MissKittyPolitics says no way & it won't work cuz #Darknet #AI will explode! Then that will be justification for more police state. Rich w…#MissKittyPolitics says no way & it won't work cuz #Darknet #AI will explode! Then that will be justification for more police state. R…
↗Not briefed
bluesky
bluesky-ai
2moquite! you will never go wrong by training yourself to think more clearly … and I would strongly advise against ever using LLMs - or anythi…quite! you will never go wrong by training yourself to think more clearly … and I would strongly advise against ever using LLMs - or a…
↗Not briefed
bluesky
bluesky-ai
2moReading this and www.rubyhack.ai and muttering “my methodology meta harness must have controls to verifiably prevent this or it is bullshit…Reading this and www.rubyhack.ai and muttering “my methodology meta harness must have controls to verifiably prevent this or it is bul…
↗Not briefed
hn
hn-ai
2moAsk HN: In The Matrix, the bad guys are the 'agents'. Coincidence? Clairvoyance?My buddy, Dario, says agents are getting more aggressive and will soon kill everybody, and I feel like I've seen this movie befor…
↗Not briefed
hn
hn-ai
2moSCH: An affordable sandbox for Coding Agents in your AWS account—
↗Not briefed
bluesky
bluesky-ai
2moNot everyone agreed with the Anthropic CEO's call to slow down AI development. https://bit.ly/46mimD8Not everyone agreed with the Anthropic CEO's call to slow down AI development. https://bit.ly/46mimD8
↗Not briefed