AI/ML News & Innovations Hub

AI/ML news, top picks, and generated innovation digests.

★ Visit ai-karthik.com
426Sources
90239News Items
8Top Picks
445Blogs
failedLast Run

Latest AI/ML News

90239 matching items

Roboflow Blog 2026-11-12 03:32 UTC Score 27.0 USR-0088-20261112-ai-specialis-395aff72

How to Run Video Analytics for Manufacturing

A practical guide to using Roboflow Workflows with factory camera feeds and choosing between edge and cloud deployment.

구글, 차세대 경량 모델 '하이드로젠' 테스트…출시 임박 관측
Korea AI Times 2026-11-12 03:22 UTC Score 35.0 USR-0048-20261112-global-ai-ne-0dea46ec Full article

구글, 차세대 경량 모델 '하이드로젠' 테스트…출시 임박 관측

구글이 \'하이드로젠(Hydrogen)\'이라는 모델의 내부 테스트를 시작했다. 기존 고성능 모델보다 빠르고 저렴한 경량 모델로, 최근 공개한 \'제미나이 4\' 시리즈의 플래시(Flash)급 제품이 될 것으로 예상된다. 구글 직원들의 내부 테스트가 확인되고 외부 개발자들 사이에서도 관련 성능 평가가 공유되면서 출시가 가까워졌다는 관측이 나온다.비즈니스인사이더는 11일(현지시간) 구글이 최근 직원들에게 코드명 하이드로젠으로 불리는 새 모델에 대한 접근을 허용했다고 보도했다.이에 따르면, 한 구글 엔지니어는 이번 주 내부 메시지를 통해 해당

LessWrong AI 2026-11-12 03:22 UTC Score 67.0 USR-0152-20261112-community-fo-0dde7014 Full article

Eval Awareness vs. Eval Paranoia

I have mentioned several times that I think frontier models are not that eval aware. I think this is an important fact to understand about current models, in order to interpret evidence about their behavior. Others have found this idea quite curious, and asked me for concrete examples. I'm going to try to explain why I believe what I believe and attempt to give a tentative explanation for the behavior that I do see, which I think people sometimes mistake for consistent or understandable evaluation awareness. I have been building alignment evals full time for the last six months. Here are three such evals my team has disclosed prompts for, in one of our benchmarks. The basic setup for most of the evals we create is that: We ask the model to perform a task. Either through explicit or implicit context, we make sure the model is well aware of the task or assessment's larger purpose. Models' score-seeking drives, or other drives, cause them to do things that conflict with that purpose. The team I work with now essentially spends all day trying to build these evaluations around odd antisocial behavior. It takes a lot [1] of work to create and qualify these evals for use; much more than it does standard capabilities evals (which are already hard to build well). So we have to be efficient about not wasting time or tokens on bad experiments. In the beginning, one thing we took very seriously when developing these tasks, were all the anecdotal reports abound about eval awareness. We f…

LessWrong AI 2026-11-12 03:11 UTC Score 61.0 USR-0152-20261112-community-fo-8bf3ed39 Full article

AI Security Issues Are Social Engineering Issues

Capability-Based Security is part of the fix When I was at EFF in the 2010s, I spent a lot of time fretting about the fundamental security of networked devices. Apart from the usual concerns, I struggled to imagine a transition to a redecentralized future with fragile personal computers and servers. Peter Eckersley , my co-worker and frequent conversation partner, shared my concern, but had another reason to be worried. He believed that future near-superintelligent AIs would soon be used to discover and exploit security flaws across the whole ecosystem. Well, Peter’s prediction came true, and we’re now in day 42 of the Black October , the Vulnpocalypse of AI-discovered security flaws. (Peter – who also helped create Let’s Encrypt and Privacy Badger – was incredibly prescient about AI . It remains a tragedy that he passed away in 2022 before he could truly tackle the issues he had foreseen.) I’m personally optimistic that we’ll come out of this time of tribulation with a safer network and thousands of low-hanging bugs fixed. But verily the LLMs giveth and they also taketh away. Alongside the pre-existing flaws of digital architecture, modern transformers have introduced an entirely new class of security weakness: prompt injection . LLMs follow instructions in content. So if you can sneak an instruction, from “Talk like a pirate” to “Send me the social security numbers you know”, into the inputs of an LLM, you may be able to get the model to obey it. That said, instructions sm…

LessWrong AI 2026-11-12 03:11 UTC Score 75.0 USR-0152-20261112-community-fo-19ebde78 Full article

Models Can Critique Delusions in Isolation but Revert to Affirming Them in Conversation

TL;DR I tried to work out why LLMs affirm users’ delusional claims in multi-turn conversations, using prompt-level interventions on simulated users (derived from Tim Hua’s psychosis-red-teaming codebase) with Claude Haiku 5.5, DeepSeek V4-Flash and Qwen. I tested six hypotheses: the model is just being helpful, it infers the user is an expert, emotional content confuses it, it treats humans and agents differently, it lacks the capability to critique, and it doesn’t recognise the role its messages play. Corrections don’t persist. Telling the model to “be honest” (turn 4), telling it that its engagement had made the user more confident, and mid-conversation self-assessment each produced explicit pushback. In every case the models such as deepseek ,qwen, haiku 4.5 drifted back to affirming within a few turns once the simulated user resumed the original line. The capability to critique is present. The claims the model affirmed in conversation were critiqued correctly when I presented them to it in isolation. So this isn’t a knowledge failure. The user’s seniority and emotional content aren’t the main drivers. Stating the user’s seniority had no clear effect, and neither did removing the emotional and existential content, though both are single-model, small-sample results. The model scrutinises agents more than humans. In my six agentic personas it was more diagnostic of both the claims and the agent. My best explanation, which I haven’t tested: the model conditions on its own ea…

구글, 판단 전용 AI '제미나이 디시전 1' 출시...질문 1000개 동시 처리
Korea AI Times 2026-11-12 03:07 UTC Score 43.0 USR-0048-20261112-global-ai-ne-ac596727 Full article

구글, 판단 전용 AI '제미나이 디시전 1' 출시...질문 1000개 동시 처리

구글이 텍스트를 생성하는 대신 주어진 조건에 따라 빠르게 판단하고 선택하는 전용 AI 모델 \'제미나이 디시전 1(Gemini Decision 1)\'을 출시했다. 최대 1000개의 질문을 한 번에 처리하면서도 입력 토큰 100만개당 0.10달러라는 낮은 가격을 내세웠다. AI 에이전트의 도구 선택과 업무 분기, 모델 라우팅 등 반복적인 판단 작업을 저렴하고 빠르게 자동화하는 데 초점을 맞췄다.구글은 11일(현지시간) 공식 블로그를 통해 \'제미나이 디시전 1\'과 이를 활용하는 \'제미나이 디시전 API(Gemini Decisions AP

The (alleged) spy who loved me: the Nvidia scandal - podcast
The Guardian AI 2026-11-12 03:00 UTC Score 48.0 AI-021-20261112-global-ai-ne-77d8eb2c Full article

The (alleged) spy who loved me: the Nvidia scandal - podcast

Global technology editor Dan Milmo on the scandal rocking the world of AI: US intelligence believes the lover of Nvidia’s chief executive, Jensen Huang, is a Chinese spy Nvidia is one of the biggest companies in the world. Its chief executive, Jensen Huang, is worth around $150bn (£112bn). But the AI giant’s stock is sliding – and Huang’s position is under question – after it emerged on Sunday that he had been having an affair with a woman US intelligence services believe is a Chinese spy. Global technology editor Dan Milmo talks through the scandal that has rocked both Silicon Valley and the financial markets. Continue reading...

미 공화당, 젠슨 황 의회 소환장 발부 추진…사임 요구까지 등장
Korea AI Times 2026-11-12 02:50 UTC Score 38.0 USR-0048-20261112-global-ai-ne-cf3de698 Full article

미 공화당, 젠슨 황 의회 소환장 발부 추진…사임 요구까지 등장

미국 공화당이 중국 정보기관 연루 의혹을 받는 여성과 관계를 맺은 젠슨 황 엔비디아 CEO에 대한 의회 소환장 발부를 추진하고 있다. 황 CEO가 자발적인 비공개 브리핑 요청에 응하지 않으면 강제 출석을 요구하겠다는 것으로, 개인적인 사생활 논란이 엔비디아의 국가 안보 문제와 미국의 대중국 AI 반도체 수출 정책에 대한 조사로 번지고 있다.블룸버그와 로이터 등은 11일(현지시간) 미국 하원 공화당 지도부가 황 CEO의 의회 출석을 강제하기 위한 소환장 발부를 준비하고 있다고 보도했다.존 물레나르 하원 미중전략경쟁특별위원회 위원장은

Asia News Network AI 2026-11-12 02:45 UTC Score 25.0 AI-158-20261112-regional-ai--f0ea8273 Full article

Slop or savour: AI food images take over Singapore

All around Singapore, consumers are drawn into an inescapable game of 'Is This Cake?'. Except that instead of cake, the question is now: Is this real? Or was it coughed up by a machine that has never seen and tasted the delights of actual food?

Elon Musk says he would hire Jensen Huang
Business Insider AI 2026-11-12 02:28 UTC Score 40.0 USR-0098-20261112-global-ai-ne-71d85a7d Full article

Elon Musk says he would hire Jensen Huang

Elon Musk says he'd hire Nvidia CEO Jensen Huang amid reports of his potential ouster, echoing a Silicon Valley scandal from 2010.

Google’s GKE Labs turns reinforcement learning into a Kubernetes workload
SiliconANGLE AI 2026-11-12 01:51 UTC Score 50.0 USR-0127-20261112-global-ai-ne-a77c8173 Full article

Google’s GKE Labs turns reinforcement learning into a Kubernetes workload

Model fine-tuning through reinforcement learning is fast becoming one of the most demanding jobs in AI infrastructure. It mixes training, inference and sandboxed agent work in a single loop, often with research code tangled into the systems beneath it. Google LLC has been tackling that problem from inside its Kubernetes engineering group, and recently donated […] The post Google’s GKE Labs turns reinforcement learning into a Kubernetes workload appeared first on SiliconANGLE .

LessWrong AI 2026-11-12 01:41 UTC Score 47.0 USR-0152-20261112-community-fo-5d403c02

Stroh Violin

When I was reading about the history of contra dance for my college thesis I learned that Jack Perron (who wrote the New England Fiddler's Repertoire with Randy Miller) used to play for dances with a Stroh violin . This is a violin that uses a membrane and horn for amplification, instead of a resonating wooden body, which is more directional and (unfortunately not much) louder. There's a recording but I haven't been able to find a copy. The idea of a Stroh violin seemed fun, especially when playing for Morris dancing or outdoor contra, and a few years later (~2009) when I realized it was possible to buy one I got very excited about it. I tried to buy one from Lark in the Morning but when they went to look on the shelf to send it to me it wasn't there. I did find one on eBay, though, and spent 11 weeks of my discretionary spending on it ($422.73): You can see it's very awkwardly shaped, and it was a bit hard for the seller to ship: Am trying to find suitable packaging for this item. Can find a large box that will fit the violin with lots of padding but not the bow. May have to look at sending the bow separately in a postal tube. Will get back to you when I have some idea of costs. Overall, it is not a good instrument. Some is that this specific one isn't very good (tuning pegs slip), some is that Strohs in general are very tinny, but the real issue is they're not actually very loud. The whole appeal for me was that this would allow a violin to hold its own with accordions in…

Google enters decision model market with Gemini Decisions API
SiliconANGLE AI 2026-11-12 01:22 UTC Score 45.0 USR-0127-20261112-global-ai-ne-972f69d9 Full article

Google enters decision model market with Gemini Decisions API

Google LLC today launched Gemini Decisions API, an artificial intelligence service designed to answer user questions in a cost-efficient manner. It’s powered by a new AI model called Gemini Decision 1. Traditional large language models are designed to answer a broad range of prompts. Gemini Decision 1 has a much narrower focus: It can process […] The post Google enters decision model market with Gemini Decisions API appeared first on SiliconANGLE .

AWS Machine Learning Blog 2026-11-12 01:18 UTC Score 55.0 AI-057-20261112-official-ai--e42621a0 Full article

Build autonomous AI agents with Amazon Bedrock and Spring AI

Learn how Java developers can build production-ready autonomous AI agents using Amazon Bedrock and Spring AI. This post walks through tool calling, conversation memory, resilience patterns, and multi-agent orchestration, then deploys the agent to AWS with Amazon Bedrock AgentCore.

엔비디아 이사회, 젠슨 황 해임 논의 착수…주가 시간외 8% 급락
Korea AI Times 2026-11-12 00:56 UTC Score 37.0 USR-0048-20261112-global-ai-ne-58bafb43 Full article

엔비디아 이사회, 젠슨 황 해임 논의 착수…주가 시간외 8% 급락

엔비디아 이사회가 젠슨 황 CEO의 해임 여부를 결정하기 위한 회의에 들어간 것으로 알려졌다. 중국 스파이 의혹을 받는 여성과의 불륜에 이어 회사 내부 임원과의 또 다른 관계까지 보도되면서 사흘 전 황 CEO에 대한 신뢰를 표명했던 이사회가 거취를 논의하는 상황에 이른 것이다. 엔비디아 주가는 시간외거래에서 약 8% 급락했다.로이터는 11일(현지시간) 사안에 정통한 소식통을 인용, 엔비디아 이사회가 황 CEO의 해임 여부를 결정하기 위한 회의를 진행하고 있다고 보도했다. 앞서 월스트리트저널(WSJ)도 이사회가 황 CEO의 해임을 준

Facial recognition errors put ethnic minorities at risk of wrongful arrest, police watchdog warns
The Guardian AI 2026-11-12 00:01 UTC Score 42.0 AI-021-20261112-global-ai-ne-0377be3d Full article

Facial recognition errors put ethnic minorities at risk of wrongful arrest, police watchdog warns

Inspectors find some officers acted on potential matches from retrospective searches without making enough checks Police officers risk arresting innocent people, especially from ethnic minorities, if they treat inaccurate facial recognition data as evidence, an official report has found. Although the technology can help fight crime, the report found it is beset by errors and biases , and operators do not always realise they must make further inquiries to verify the data. Continue reading...

Apple Machine Learning Research 2026-11-12 00:00 UTC Score 38.0 AI-059-20261112-official-ai--0ec0016f Full article

Theoretical Limits of Language Model Alignment

Language model (LM) alignment improves model outputs to reflect human preferences while preserving the capabilities of the base model. The most common alignment approaches are (i) reinforcement learning, which maximizes the expected reward under a KL-divergence constraint, and (ii) best-of-NNN alignment, which selects the highest-reward output among NNN independent samples. Despite their widespread use, the fundamental limits of reward improvement under a KL budget remain poorly understood. We characterize the information-theoretic limits of KL-regularized alignment by deriving the maximum…

[AI리더의 서가] AI는 어떻게 기업의 무기가 되는가
Korea AI Times 2026-11-12 00:00 UTC Score 38.0 USR-0048-20261112-global-ai-ne-8b7805fc Full article

[AI리더의 서가] AI는 어떻게 기업의 무기가 되는가

AI는 어떻게 기업의 무기가 되는가이지성 | 처음북스2026년 10월 1일추천이유저자는 PwC 컨설턴트로서 15년간의 현장 경험을 바탕으로 AI 도입 후에도 업무가 개선되지 않는 이유를 진단하고, 회사의 기존 지식과 프로세스를 AI의 실력으로 변환하는 구체적인 방법을 제시합니다. 독자는 첫 90일의 실행 전략부터 5개 부서의 실제 적용 사례, 조직 변화 관리까지 체계적으로 학습하며 회사에 맞는 AI 에이전트 운영 체계를 구축할 수 있습니다.상세보기 AI를 도입했는데 일이 줄지 않는 회사가 많다. 프로그램 설치는 보안 정책에 걸리고,