How we detect frontier model misalignment
OpenAI uses layered monitoring to detect frontier model misalignment and unauthorized agent behavior so responders can intervene before incidents escalate.
AI와 IT의 변화를 이해하고, 실제 서비스에 적용하는 데 필요한 정보를 모읍니다.
OpenAI uses layered monitoring to detect frontier model misalignment and unauthorized agent behavior so responders can intervene before incidents escalate.
The post We’re redefining the SDLC for the agentic era. Lots of GitHub updates this morning at Universe [Read more] appeared first on Source .
The acquisition would bring two major players in on-demand delivery under one roof, combining DoorDash’s dominance in restaurant takeout with Instacart’s reach in grocery delivery.
On Thursday, AI app Wabi, now describing itself as a "personal agent," said it has raised $55 million at a $1.25 billion valuation.
The social networking site is facing criticism for blocking access to the site, which identifies thousands of ICE agents and other employees working for Homeland Security.
<blockquote cite="https://twitter.com/benhylak/status/2115172657910120740"><p>every product i use should just be a headless mcp, that feeds into an agent, where i never have to think about anything underneath.</p> <p>except for our product, of course. </p> <p>in our product, cust
안녕하세요. ML 연구와 엔지니어링 작업에 사용하고 있는 Claude Code 하네스, oh-my-ouroboros를 공개합니다. https://github.com/DaeSeokSong/oh-my-ouroboros Claude Code를 오래 사용하다 보면 규칙과 스킬이 계속 늘어납니다. 같은 실수를 반복하지 않게 규칙을 추가하지만, 시간이 지나면 어떤 규
안녕하세요. Specify를 개발하고 있습니다. 지난 1년 동안 Claude Code와 Codex로 프로덕트를 만들면서, 기능이 동작하고 테스트를 통과해도 프로덕트의 구조와 설계 의도에 맞는지는 별도로 확인해야 한다는 고민이 생겼습니다. 에이전트가 만든 코드가 빠르게 쌓일수록 데이터 흐름과 기능 간 관계를 전체적
Granite Switch 4.1.1 — Update models like you update software
이스라엘 광고 기술 기업 Discover-Tech 가 Starlink 사용자에게 대규모로 광고를 노출했으며, 위치 파악이나 감시가 목적이었을 가능성이 크다고 TrendAI 연구진이 판단함 광고 데이터 분석 결과, 이 업체는 100개국 이상에서 Starlink 트래픽 대상 광고 입찰 1위 였...
Comments
Pwn2Own Berlin 2026을 위해 메모리 손상 없이 논리 버그만으로 Microsoft Edge에서 임의 코드 실행까지 이어지는 체인을 구축함 브라우저 내부 탐색 API가 출발 탭을 명시하지 않으면 활성 탭을 사용 하는 동작을 이용해, 다른 사이트에서 JavaScript를 실행하는 범...
Comments
Comments
Comments
Mistral이 1조 파라미터 네이티브 멀티모달 모델 Mistral Large 4 의 개선된 가중치를 공개함. 공개 프리뷰 이후 강화학습을 확대해 코딩, 방어적 사이버보안, 추론 효율을 높임 Artificial Analysis Coding Agent Index 에서 GLM-5.3와 사실상 동률인 48%로 오픈웨이...
Mintive was founded by two engineers who have been working on AI for decades through Google Brain and Coursera. It's platform promises to help IT easily control enterprise agents.
The New York Times is testing a new AI feature for its Cooking app that helps users to modify recipes, and it has sparked some drama among users. Last Friday, NYT Cooking senior editor Genevieve Ko announced the feature, which can be used to recommend substitutions for ingredient
Over the last year we’ve been sharing our work leveraging large language models to catch bugs on the fly — a novel approach we call Catching Just-in-Time Tests (JiTTests). Now, we’re sharing the results and positive impact we’ve seen with Catching JiTTests deployed into productio
Google's negative cash flow balloons to -$9.06 billion, and stock price tumbles following Q3 earnings.
When it was revealed earlier this month that Withings would require new users to subscribe to a Withings Plus subscription to sync heath data from the company's devices to third-party apps, Tom Anthony, who was already fed up with the Withings mobile app which they describe as a
NVIDIA Nemotron 3.5 Super VL is a 120B-parameter mixture-of-experts model that pairs reasoning and tool use with multimodal understanding. This post explores its architecture and use cases, then shows how to deploy it on Amazon EKS, EKS Auto Mode, and Amazon SageMaker HyperPod EK
SuperMe is a network of AI profiles built from the knowledge of real-life experts. When you ask a question, the platform determines which experts are best suited to answer and creates a group chat with their AI profiles.
Dream is grounded in real physics and construction standards. Just as LLMs learn from text, Dream's world model is trained on more than 100 million square feet of building data gathered from permit archives across the United States.
To anyone listening closely to what Microsoft CEO Satya Nadella has been saying over the past couple of years, this week felt like the clearest sign yet that AI is about to fundamentally change the size of Microsoft's workforce. Microsoft is now offering voluntary buyouts to arou
해킹 단체 Department of Peace 가 미국 국토안보부(DHS) 인사 파일을 탈취해 요원과 관계자 4만 명 이상의 이름, 전화번호, 주소를 공개했으며, 잠복 작전 관련 인력도 포함됨 해커들은 DHS의 통합 로그인(SSO) 시스템 을 무력화해 데이터를 훔쳤다고 밝혔으며, 자료...
Comments
The US power sector could release 42 percent more planet-heating carbon dioxide pollution by 2030 than it did in 2025 as it builds out more fossil fuel infrastructure to meet data center electricity demand, according to a new analysis by the Sierra Club. That would essentially wi
A top executive of a Singapore AI cloud company that was accused of illegally selling Nvidia chips to China could also have been a Chinese spy, sources have told the New York Times.
Mistral says its new model can afford organizations a kind of control over its cyber defenses that closed models can't.