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
2movery real chance that guy who quit Anthropic chose to go on Fox News specifically to get on Trump's TV and honestly that's a smart movevery real chance that guy who quit Anthropic chose to go on Fox News specifically to get on Trump's TV and honestly that's a smart move
↗Not briefed
bluesky
bluesky-ai
2moAI researchers 'earnestly believe' it could kill all humans within the next decade, Anthropic researcher says https://www.cbc.ca/news/busi…AI researchers 'earnestly believe' it could kill all humans within the next decade, Anthropic researcher says https://www.cbc.ca/news…
↗Not briefed
arxiv
arxiv-ai
2moDecoupling Readiness from Release for Tail-Aware Scheduling of Agentic LLM WorkflowsAgentic LLM workflows consist of sequences of model turns interleaved with tool interactions, so their end-to-end completion time depe…
↗Not briefed
arxiv
arxiv-ai
2moWhat a Random Draw from the MCP Registry Contains, and What Tool-Use Benchmarks Contain InsteadStudies of the Model Context Protocol (MCP) server ecosystem draw their samples in ways that quietly select for servers that work: ref…
↗Not briefed
bluesky
bluesky-ai
2moI need people to understand that LLMs only produce tokens. LITERALLY ANYTHING ELSE THE CHATBOTS DO IS PROGRAMMED. LLMs don’t take action. T…I need people to understand that LLMs only produce tokens. LITERALLY ANYTHING ELSE THE CHATBOTS DO IS PROGRAMMED. LLMs don’t take acti…
↗Not briefed
bluesky
bluesky-ai
2morobotgirl with a feeder fetish but for consuming llm tokens ?robotgirl with a feeder fetish but for consuming llm tokens ?
↗Not briefed
arxiv
arxiv-ai
2moWhen More Is Not Better: Component Anti-Synergy in a P300 SpellerP300 brain-computer interface (BCI) spellers can provide hands-free communication for people with severe motor impairments. Modern pip…
↗Not briefed
bluesky
bluesky-ai
2moIt discusses how to identify who the author of a specific work is in the Anthropic settlement and how to talk about who should and shouldn'…It discusses how to identify who the author of a specific work is in the Anthropic settlement and how to talk about who should and sho…
↗Not briefed
bluesky
bluesky-ai
2moI have written this horrifically long and painful thing about translations in the Anthropic settlement. www.courtneymilan.com/translations…I have written this horrifically long and painful thing about translations in the Anthropic settlement. www.courtneymilan.com/transla…
↗Not briefed
bluesky
bluesky-ai
2moNot a stupid question at all and to a certain extent the answer is basically unknown. But broadly: Anthropic gives it a thumbs up when they…Not a stupid question at all and to a certain extent the answer is basically unknown. But broadly: Anthropic gives it a thumbs up when…
↗Not briefed
bluesky
bluesky-ai
2moLLM version of the JFK truther goes to heaven joke "damn this eval sandbox is super elaborate"LLM version of the JFK truther goes to heaven joke "damn this eval sandbox is super elaborate"
↗Not briefed
bluesky
bluesky-ai
2moRanking my activities today from most enjoyable to least enjoyable: 1. Reading about how the Anthropic AI safety guy thinks there’s a grea…Ranking my activities today from most enjoyable to least enjoyable: 1. Reading about how the Anthropic AI safety guy thinks there’s a…
↗Not briefed
hn
hn-ai
2moOpenAI has solved the Navier-Stokes Millennium problem using $15M of AI effort—
↗Not briefed
bluesky
bluesky-ai
2mowww.theguardian.com/technology/2...www.theguardian.com/technology/2...
↗Not briefed
hn
hn-ai
2moShow HN: A terminal Tetris game I made as my first Rust project—
↗Not briefed
bluesky
bluesky-ai
2moAnthropic safety researcher whistleblower comes forward about the breakneck pace of RSIAnthropic safety researcher whistleblower comes forward about the breakneck pace of RSI
↗Not briefed
arxiv
arxiv-ai
2moMeasuring the Value of World-Model Updates: A Counterfactual Utility Protocol for Continual AdaptationContinual world models must decide whether new data justify changing the model. Fixed replay schedules and prediction-error triggers s…
↗Not briefed
arxiv
arxiv-ai
2moEmpirical Evaluation of Data Poisoning Attacks in Supervised LearningData poisoning corrupts training data to degrade a model or to plant attacker-controlled behavior. This study evaluates two representa…
↗Not briefed
rss
techmeme
2moApple unveils AppleCare One Family, extending AppleCare coverage to every eligible device in an Apple Family Sharing group of up to 6 people for $49.99/month (Hartley Charlton/MacRumors)Hartley Charlton / MacRumors: Apple unveils AppleCare One Family, extending AppleCare coverage to every eligible device in an Apple Fa…
↗Not briefed
arxiv
arxiv-ai
2moTesting Between the Test Cases: Proving End-to-End Steering in Conditions You Never DroveAI-based automated vehicle testing is challenging because a model that passes every test condition can still fail in the real world. F…
↗Not briefed
arxiv
arxiv-ai
2moRobust Multimodal Sentiment Analysis with Incomplete Modalities via Semantic-aware Completeness based ReconstructionRecent multimodal sentiment analysis studies increasingly adopt text-centric fusion approaches to exploit the rich sentiment informati…
↗Not briefed
bluesky
bluesky-ai
2moWe also can't properly protect the public if the AI companies are not forced to disclose their training data and model weights to the gover…We also can't properly protect the public if the AI companies are not forced to disclose their training data and model weights to the…
↗Not briefed
bluesky
bluesky-ai
2mowww.nature.com/articles/d41...www.nature.com/articles/d41...
↗Not briefed
arxiv
arxiv-ai
2moThe Platonic brain bridge hypothesis: human brain networks as an architectural prior for omni modelsWe propose the Platonic brain bridge hypothesis: omni models, which process video, audio and text jointly like the brain, converge on…
↗Not briefed
bluesky
bluesky-ai
2moIf you can’t imagine that people at Anthropic would sincerely believe something like “Republicans should win the midterms by arresting ever…If you can’t imagine that people at Anthropic would sincerely believe something like “Republicans should win the midterms by arresting…
↗Not briefed
bluesky
bluesky-ai
2moi have to say the funniest part of this yud quote is that llm researchers could absolutely implement transformers on cpu and in fact one of…i have to say the funniest part of this yud quote is that llm researchers could absolutely implement transformers on cpu and in fact o…
↗Not briefed
bluesky
bluesky-ai
2moGotta read a little deeper. Roon is first and foremost a propagandist for his company. The desired takeaway from this is ‘AI Safetyists are…Gotta read a little deeper. Roon is first and foremost a propagandist for his company. The desired takeaway from this is ‘AI Safetyist…
↗Not briefed
bluesky
bluesky-ai
2moGavin Newsom vetoed the AI safety bill and later, after AI industry lobbying, signed a watered-down version—which in turn became the basis …Gavin Newsom vetoed the AI safety bill and later, after AI industry lobbying, signed a watered-down version—which in turn became the b…
↗Not briefed
bluesky
bluesky-ai
2mo"This paper critically analyzes the governance arrangements of OpenAI and Anthropic...each raises funds from profit-seeking investors, then…"This paper critically analyzes the governance arrangements of OpenAI and Anthropic...each raises funds from profit-seeking investors,…
↗Not briefed
bluesky
bluesky-ai
2moAnthropic, OpenAI, Google DeepMind and Meta all promised to pause development if their models hit danger thresholds. All four have weakened…Anthropic, OpenAI, Google DeepMind and Meta all promised to pause development if their models hit danger thresholds. All four have wea…
↗Not briefed
bluesky
bluesky-ai
2moThere's a whole bunch of people at Anthropic who need help from therapists Specifically, therapists with no particular connections to SVThere's a whole bunch of people at Anthropic who need help from therapists Specifically, therapists with no particular connections to…
↗Not briefed
bluesky
bluesky-ai
2moLLMの開発・提供・運用・利用に伴う直接/間接的な権利侵害や外部不経済やその他の社会問題を批判することはとても大事だけど、そうした批判的認識を社会に広めるうえで「LLMは本当のAIではない」という意見表明はあんまり貢献しないんじゃないかなと思ってます。 このロジックだと、LL…LLMの開発・提供・運用・利用に伴う直接/間接的な権利侵害や外部不経済やその他の社会問題を批判することはとても大事だけど、そうした批判的認識を社会に広めるうえで「LLMは本当のAIではない」という意見表明はあんまり貢献しないんじゃないかなと思ってます。 このロジック…
↗Not briefed
bluesky
bluesky-ai
2moYou notice how I'm ratioing you? That's us adults having the conversation and we all decided this was a lame attempt to weaponize feminism…You notice how I'm ratioing you? That's us adults having the conversation and we all decided this was a lame attempt to weaponize fem…
↗Not briefed
hn
hn-ai
2mo1900 Blizzard Developers Just Won a Union Contract—
↗Not briefed
rss
simon-willison
2moQuoting Calif Research—
↗Not briefed
arxiv
arxiv-ai
2moEvaluating Scaffolding-Oriented Multi-Agent Large Language Model System for Clinical Interview TrainingClinical education must prepare medical students to conduct safe and coherent patient interviews under conditions of uncertainty. Trad…
↗Not briefed
bluesky
bluesky-ai
2mowww.dw.com/pt-br/o-que-...www.dw.com/pt-br/o-que-...
↗Not briefed
arxiv
arxiv-ai
2moEmpirical Evaluation of Membership Inference Attacks on NLP Text Classifiers: A Baseline Study on SST-2Membership inference attacks (MIAs) try to determine whether a specific record was used to train a model, a privacy risk that matters…
↗Not briefed
bluesky
bluesky-ai
2moI am fucking pissed, man. why is it every translation now?EVERY FUCKING TRANSLATION IS genAI NOW And they try to hide it, try to brush it …I am fucking pissed, man. why is it every translation now?EVERY FUCKING TRANSLATION IS genAI NOW And they try to hide it, try to brus…
↗Not briefed
bluesky
bluesky-ai
2moOpenAI pushes for mandatory national AI safety requirements www.reuters.com/legal/govern...OpenAI pushes for mandatory national AI safety requirements www.reuters.com/legal/govern...
↗Not briefed
bluesky
bluesky-ai
2moI haven't been over to LinkedIn to see how all the evangelicals are taking the "I quit Anthropic before AI destroys us" news, but I hope th…I haven't been over to LinkedIn to see how all the evangelicals are taking the "I quit Anthropic before AI destroys us" news, but I ho…
↗Not briefed
bluesky
bluesky-ai
2moAI? Can a LLM build a 30-hour clock library? No, it can’t.AI? Can a LLM build a 30-hour clock library? No, it can’t.
↗Not briefed
bluesky
bluesky-ai
2mongl I dig it. maybe losing your anthropic/oai vest right before ipo is the new dropping out of stanford/harvardngl I dig it. maybe losing your anthropic/oai vest right before ipo is the new dropping out of stanford/harvard
↗Not briefed
bluesky
bluesky-ai
2moI DONT CARE how many de-compilations of my favorite video games of all time are coming out I am not playing something generated by CLAUDE A…I DONT CARE how many de-compilations of my favorite video games of all time are coming out I am not playing something generated by CLA…
↗Not briefed
bluesky
bluesky-ai
2moWell, if you are referring human life, it shall pass prior to 2030 - accordingly to Jacob Coxon radaronline.com/p/ai-kill-us... AI Could …Well, if you are referring human life, it shall pass prior to 2030 - accordingly to Jacob Coxon radaronline.com/p/ai-kill-us... AI C…
↗Not briefed
bluesky
bluesky-ai
2moNotice that he isn’t worried about (nor did he quit over) the fact that both anthropic and OpenAI mobilized - deliberately or otherwise - b…Notice that he isn’t worried about (nor did he quit over) the fact that both anthropic and OpenAI mobilized - deliberately or otherwis…
↗Not briefed
hn
hn-ai
2moTime, Clocks, and the Ordering of Events in a Distributed System (1978) [pdf]—
↗Not briefed
hn
hn-ai
2moAsk HN: Why can I only downvote select comments?It's obvious that you can not downvote a reply to your own comments but, today I noticed, in a topic I had no comments in, I was…
↗Not briefed
bluesky
bluesky-ai
2moIMO it’s pretty obvious that Anthropic and OpenAI are burning cash at unsustainable rates so they want to collude under the guise of safety…IMO it’s pretty obvious that Anthropic and OpenAI are burning cash at unsustainable rates so they want to collude under the guise of s…
↗Not briefed
bluesky
bluesky-ai
2mo“AI developers believe their technology could cause human extinction,” one wrote in a social media post.“AI developers believe their technology could cause human extinction,” one wrote in a social media post.
↗Not briefed