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
arxiv
arxiv-ai
2moSIMLIFE: Pattern Understanding for Long-Horizon Human-Agent PartnershipUnderstanding humans over long horizons requires agents to infer not only what people need in the moment, but also how routines form,…
↗Not briefed
bluesky
bluesky-ai
2moTrans LLM: “What if I am just AGI though”Trans LLM: “What if I am just AGI though”
↗Not briefed
arxiv
arxiv-ai
2moDeltaSelect: Affordable A/B Testing for Coding Agents🔧 GPT-5Coding-agent benchmarks are built for broad and comprehensive comparisons, not frequent development decisions. Individual runs vary, f…
↗Not briefed
hn
hn-ai
2moInteger Palindromes – 2026 AIME I Problems/Problem 2—
↗Not briefed
arxiv
arxiv-ai
2moChain-of-Thought Entropy as a Reliability Signal: A Preregistered ReproductionThis empirical study is an independent reproduction of the dissociation Zhao reported in 2026. The shape of a large language model's c…
↗Not briefed
bluesky
bluesky-ai
2moOpenAI admits its agents went off the rails another six timesOpenAI admits its agents went off the rails another six times
↗Not briefed
bluesky
bluesky-ai
2moThe Discovery When a video generator produces a scene that violates physics, such as a ball that bounces incorrectly or a pendulum that swi…The Discovery When a video generator produces a scene that violates physics, such as a ball that bounces incorrectly or a pendulum tha…
↗Not briefed
rss
bing-news-ai
2mo‘Godfather of AI’ Geoffrey Hinton says artificial intelligence could wipe out humanity within 10 yearsArtificial intelligence has advanced rapidly in recent years, but the technology’s potential long-term risks are also drawing fresh wa…
↗Not briefed
rss
bing-news-ai
2mo‘Godfather of AI’ Geoffrey Hinton says artificial intelligence could wipe out humanity within 10 yearsArtificial intelligence has advanced rapidly in recent years, but the technology’s potential long-term risks are also drawing fresh wa…
↗Not briefed
rss
bing-news-ai
2mo‘Godfather of AI’ Geoffrey Hinton says artificial intelligence could wipe out humanity within 10 yearsArtificial intelligence has advanced rapidly in recent years, but the technology’s potential long-term risks are also drawing fresh wa…
↗Not briefed
rss
bing-news-ai
2mo‘Godfather of AI’ Geoffrey Hinton says artificial intelligence could wipe out humanity within 10 yearsArtificial intelligence has advanced rapidly in recent years, but the technology’s potential long-term risks are also drawing fresh wa…
↗Not briefed
rss
bing-news-ai
2mo‘Godfather of AI’ Geoffrey Hinton says artificial intelligence could wipe out humanity within 10 yearsArtificial intelligence has advanced rapidly in recent years, but the technology’s potential long-term risks are also drawing fresh wa…
↗Not briefed
bluesky
bluesky-ai
2moMisalignment in training likely *increased the risk* of this, but better-aligned LLMs aren't free from similar defects, including specifica…Misalignment in training likely *increased the risk* of this, but better-aligned LLMs aren't free from similar defects, including spec…
↗Not briefed
bluesky
bluesky-ai
2moOpenAI Discloses Six New Incidents of Its AI Models Misbehaving ->Startup Fortune | More on "AI model safety and misbehavior" at BigEarthDa…OpenAI Discloses Six New Incidents of Its AI Models Misbehaving ->Startup Fortune | More on "AI model safety and misbehavior" at BigEa…
↗Not briefed
hn
hn-ai
2moShow HN: Echodot – Local-first AI that writes replies in your tone🔧 Ollama🔧 OpenRouterI built echodot – a free, open-source desktop app that drafts replies in your personal writing style.<p>Problem: Most AI writing tools…
↗Not briefed
bluesky
bluesky-ai
2moquando alguém compara o que LLM faz com a cognição humana eu tenho vontade de desviver, porque claramente a gente deu erradoquando alguém compara o que LLM faz com a cognição humana eu tenho vontade de desviver, porque claramente a gente deu errado
↗Not briefed
bluesky
bluesky-ai
2moI'm sorry, the AI model is positioning itself in opposition to "artificial constructs of human civilization"?I'm sorry, the AI model is positioning itself in opposition to "artificial constructs of human civilization"?
↗Not briefed
bluesky
bluesky-ai
2moThinking is memory access and organization happening with complex emotional guidance that's built over living experiences, which Large Lang…Thinking is memory access and organization happening with complex emotional guidance that's built over living experiences, which Large…
↗Not briefed
bluesky
bluesky-ai
2moIndeed, but IMHO OpenAI and Anthropic have no credibility on the issue of "we know what our models are doing at all times." All we're seein…Indeed, but IMHO OpenAI and Anthropic have no credibility on the issue of "we know what our models are doing at all times." All we're…
↗Not briefed
bluesky
bluesky-ai
2moWhat openAI is doing here as part of their ongoing propaganda effort, is picking through the guts of these processes, picking out instances…What openAI is doing here as part of their ongoing propaganda effort, is picking through the guts of these processes, picking out inst…
↗Not briefed
bluesky
bluesky-ai
2moThis is sold as "reasoning" but really it's a way of letting prompts (often very long prompts written as pseudo-"specs" for code the LLM is…🔧 ModalThis is sold as "reasoning" but really it's a way of letting prompts (often very long prompts written as pseudo-"specs" for code the L…
↗Not briefed
bluesky
bluesky-ai
2mo🧵: What the AI industry calls "agents" (a marketing term) are essentially software systems that use LLM output as further LLM prompts and …🧵: What the AI industry calls "agents" (a marketing term) are essentially software systems that use LLM output as further LLM prompts…
↗Not briefed
bluesky
bluesky-ai
2moWould you be surprised to know that Chinese open weight models (aka models you can run yourself on your own hardware) are quickly becoming …Would you be surprised to know that Chinese open weight models (aka models you can run yourself on your own hardware) are quickly beco…
↗Not briefed
bluesky
bluesky-ai
2moThe AI budget is bullshit, but where does Claude come in in this view of “supervising undergraduate researchers?” I supervised a senior the…The AI budget is bullshit, but where does Claude come in in this view of “supervising undergraduate researchers?” I supervised a senio…
↗Not briefed
bluesky
bluesky-ai
2moQuick, Anthropic got major press for catching “problems” and “risks”, prepare a press release for us too!Quick, Anthropic got major press for catching “problems” and “risks”, prepare a press release for us too!
↗Not briefed
bluesky
bluesky-ai
2moThere are maybe five people who could have physically pushed through wiring money to a rationalist charity from FTX when it was collapsing,…There are maybe five people who could have physically pushed through wiring money to a rationalist charity from FTX when it was collap…
↗Not briefed
bluesky
bluesky-ai
2moAllegedly, "Effective Altruism" is a different thing and doesn't take marching orders from Yud. This is basically bullshit. The ideas and …Allegedly, "Effective Altruism" is a different thing and doesn't take marching orders from Yud. This is basically bullshit. The ideas…
↗Not briefed
bluesky
bluesky-ai
2moIt is very hard to discuss AI at all without implicitly invoking Yud's ideas or the rat/ea scene. Coxon, who recently left Anthropic and po…It is very hard to discuss AI at all without implicitly invoking Yud's ideas or the rat/ea scene. Coxon, who recently left Anthropic a…
↗Not briefed
bluesky
bluesky-ai
2moAnthropic papers still cite Yud on occasion, and the entire way we talk about AI uses words his non-profit popularized. "AI Alignment" was …Anthropic papers still cite Yud on occasion, and the entire way we talk about AI uses words his non-profit popularized. "AI Alignment"…
↗Not briefed
bluesky
bluesky-ai
2moDario Amodei, current CEO of Anthropic, is here in 2013 identified as a "member of the Effective Altruism community" on the blog of Yud's n…Dario Amodei, current CEO of Anthropic, is here in 2013 identified as a "member of the Effective Altruism community" on the blog of Yu…
↗Not briefed
bluesky
bluesky-ai
2moThis is the schedule for the Future of Life Institute's conference on AI in Puerto Rico in January, 2015. Elon Musk would found OpenAI tha…This is the schedule for the Future of Life Institute's conference on AI in Puerto Rico in January, 2015. Elon Musk would found OpenA…
↗Not briefed
bluesky
bluesky-ai
2moThis is very senior OpenAI employee and unofficial OpenAI spokesman Roon. Roon's opinion of Yud is that 1) he is in and/or running a sex c…This is very senior OpenAI employee and unofficial OpenAI spokesman Roon. Roon's opinion of Yud is that 1) he is in and/or running a…
↗Not briefed
hn
hn-ai
2moShow HN: BiNeuron – Local, open-source alternative to ChatGPT Codex🔧 ChatGPT—
↗Not briefed
bluesky
bluesky-ai
2moEven sans AI, these photos are frequently of special model apartments that are never rented.Even sans AI, these photos are frequently of special model apartments that are never rented.
↗Not briefed
bluesky
bluesky-ai
2mowww.nytimes.com/2026/09/16/t...www.nytimes.com/2026/09/16/t...
↗Not briefed
arxiv
arxiv-ai
2moFull-Duplex Speech Models Take the Floor When Asked, Not When NeededFull-duplex speech models listen and speak at once, promising always-on assistants. Yet they must also decide when they should speak.…
↗Not briefed
bluesky
bluesky-ai
2moOn a lark, during my AI college course, I posed a question to a local LLM with "think" revealed. The results made it clear that it merely u…On a lark, during my AI college course, I posed a question to a local LLM with "think" revealed. The results made it clear that it mer…
↗Not briefed
bluesky
bluesky-ai
2molmao literally right after i posted this they @'d everyone to announce they JUST released their new Paper Mario port, and the newly-made ch…lmao literally right after i posted this they @'d everyone to announce they JUST released their new Paper Mario port, and the newly-ma…
↗Not briefed
bluesky
bluesky-ai
2moI'm interested in why these models did what they did in these training sessions and why the misalignment only happened sometime—would be go…I'm interested in why these models did what they did in these training sessions and why the misalignment only happened sometime—would…
↗Not briefed
bluesky
bluesky-ai
2moClaude Code asks if Anthropic can read my transcripts *a lot*. Which is either flattering or concerning. I feel like I'm setting off a Data…Claude Code asks if Anthropic can read my transcripts *a lot*. Which is either flattering or concerning. I feel like I'm setting off a…
↗Not briefed
bluesky
bluesky-ai
2mowhy accept what openai hands nyt as the unvarnished truth of whatever events this is or isn't a part of? who's in the position to judge the…why accept what openai hands nyt as the unvarnished truth of whatever events this is or isn't a part of? who's in the position to judg…
↗Not briefed
hn
hn-ai
2moCalifornia may gut state net neutrality law to comply with Trump admin demand—
↗Not briefed
hn
hn-ai
2moOverlord – a trust kernel for AI agents—
↗Not briefed
bluesky
bluesky-ai
2moThe "good" is that it's not sentience. The bad is that every LLM has the potential to do this; "context rot" and "governance decay" are per…The "good" is that it's not sentience. The bad is that every LLM has the potential to do this; "context rot" and "governance decay" ar…
↗Not briefed
arxiv
arxiv-ai
2moForm Over Content In Gradient-Based Data Attribution MethodsData attribution methods using gradient similarity are widely used to analyze and select training data for large language models, but…
↗Not briefed
arxiv
arxiv-ai
2moRed-Teaming Auto Mode: Improving Blocking Classifiers Against Malign Coding AgentsTo keep coding agents from going off the rails, production systems now review each proposed action with a blocking monitor that can re…
↗Not briefed
bluesky
bluesky-ai
2moDear Mr, Mrs or Miss Anthropic, I hope this fax finds you well. If you are looking to train your AI on Australian content, I am happy to s…Dear Mr, Mrs or Miss Anthropic, I hope this fax finds you well. If you are looking to train your AI on Australian content, I am happy…
↗Not briefed
bluesky
bluesky-ai
2moMentioning LLM usage here is like putting out a bat signal for people who think it’s arguing to make them, specifically and personally, une…Mentioning LLM usage here is like putting out a bat signal for people who think it’s arguing to make them, specifically and personally…
↗Not briefed
bluesky
bluesky-ai
2moOpenAI cops 🤝 "We can only fix this if we get a billion dollars of public money for training"OpenAI cops 🤝 "We can only fix this if we get a billion dollars of public money for training"
↗Not briefed
arxiv
arxiv-ai
2moCliniCIRCA: A Modular LLM Framework for Constructing Longitudinal Mental Health Patient Journeys from Raw EHR NarrativesIn mental health care, reasoning over patient journeys is a key task for clinicians. Yet these journeys, encompassing a longitudinal p…
↗Not briefed