自动生成 | 数据来源：OpenAI, Anthropic, HuggingFace, GitHub, Apple, NVIDIA, xAI, Simon Willison, THE DECODER, ITHOME, HackerNews, ArXiv, KOL Twitter
Try ChatGPT (opens in a new window) Login. OpenAI News | OpenAI. All. Company · Research · Product · Safety · Engineering · Security · Global Affairs
Better memory for a more helpful ChatGPT. Dreaming: Better memory for a more helpful ChatGPT. Research Jun 4, 2026.
Introducing GPT-5.5, our smartest model yet—faster, more capable, and built for complex tasks like coding, research, and data analysis
Today we're launching Claude Fable 5: a Mythos-class model that we've made safe for general use.
The US government has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national.
We're launching Claude Corps, a national fellowship program for people early in their careers who are passionate about extending the frontier of AI safely.
Anthropic opens Seoul and announces new partnerships across the Korean AI ecosystem—with the enterprises, startups, and researchers behind Korea's AI future.
Our latest model, Claude Opus 4.8, is an upgrade to our Opus class of models, with stronger performance across coding, agentic tasks, and reasoning.
Powered by Apple Intelligence, the new version of Siri is profoundly more capable and conversational, and deeply integrated across products.
Next-generation Apple Intelligence and Siri AI bring helpful features to iOS 27, iPadOS 27, macOS Golden Gate, watchOS 27, and visionOS 27.
NVIDIA XR AI is now available in public beta, giving developers a framework for building multimodal AI agents for AR glasses and XR devices.
Battle for a chicken dinner with an NVIDIA ACE-powered AI teammate in Ally Duo Mode available in PUBG Arcade until June 30th.
Discover our AI Control Roadmap: a defense-in-depth system to securely manage advanced, potentially misaligned AI agents.
Google DeepMind is working alongside the UK government to co-develop an AI-powered prototype to help cut application decision times by 50%.
For the past decade, we've focused on making individual AI models more capable, helpful and safe. Today, Google DeepMind — together with partners — invests in multi-agent AI safety research.
Bringing open intelligence to all, our latest models expand context length, add support across eight languages, and include Meta Llama 3.1.
Today at Connect 2024, we shared updates for Meta AI features and released Llama 3.2, a collection of models that includes new vision capabilities.
Meta is building personal superintelligence for everyone. Explore Meta AI, our latest model Muse Spark, AI research, and tools like Vibes for AI video.
Copilot Cowork is now generally available worldwide, bringing secure, AI-powered automation for complex enterprise tasks in Microsoft 365.
The two most important elements in any AI solution are Intelligence + Trust.
Planner Agent in Microsoft 365 Copilot is now rolling out to general availability, bringing intelligent, AI-powered work and task management.
AI research from Microsoft shows that large‑scale LLMs use far less energy and water per query, improving efficiency and reducing environmental impact.
I would very much like to enjoy HN the way I did years ago, as a place where I'd discover things that I never otherwise would have come across.
The increasing AI/LLM do
I'm a software engineer with a solid full-stack background and web development. With all the noise around LLMs and AI, I’m undecided between two paths:
1. Invest time in lea
Hi HN,
ArtificialAnalysis.ai provides objective benchmarks and analysis of LLM AI models and API hosting providers so you can compare which to use in your next (or current) projec
Yann LeCun, founder of AMI Labs, called Elon Musk's xAI a 'failure,' adding that he expects it won't be able to compete with OpenAI and Anthropic. LeCun also said AI labs are riski
LeCun told CNBC that Musk's AI lab, xAI, was 'kind of a failure.' 'Elon is now in a position where it's very, very difficult for him to hire top people in AI, because he's not beha
President Trump said negotiations with Anthropic over restoring access to the company's latest AI models were 'going fine.' OpenAI CEO Sam Altman, Trump, and Google DeepMind CEO De
LeCun took a shot at how Musk runs his company. LeCun's AMI Labs recently raised $1 billion to work on 'world models' which he sees as key to the next stage of AI.
Unreal Engine 5.8 adds an open-standard plugin for LLM systems, allowing developers to connect any LLM to understand the engine and projects for AI-assisted asset creation, code, a
Reinforcement learning (RL) has become a cornerstone for fine-tuning large language models (LLM). This paper rethinks the trust region approach in RL for LLM training.
Researchers at Google DeepMind discuss the progression from artificial general intelligence (AGI) to artificial superintelligence (ASI).
This paper proposes methods for detecting AI-generated text in educational content using machine learning and explainable AI for academic integrity.
We propose GRESO (GRPO with Efficient Selective Rollout), an online, lightweight pre-rollout filtering algorithm that predicts and skips uninformative prompts for LLM reasoning.
Test Input Generators (TIGs) are crucial to assess deep learning models. This paper benchmarks generative AI models for automated test input generation.
Lower prior trust in human agents (priests, peers, adults) consistently predicted higher AI selection, supporting deferred trust as a mechanism for AI adoption.
Multimodal machine learning is a vibrant multi-disciplinary research field that aims to design computer agents with intelligent multimodal capabilities.
Paper on 'Constrained Behavior Tree Generation for Safe LLM-Driven Robot Navigation' presented at IEEE EIT 2026, focusing on improving safety and reliability of AI-controlled robot
Probably raises $9M in seed funding led by Andreessen Horowitz to develop AI tools that eliminate hallucinations and improve accuracy through deterministic validation systems for L
Perspective article comparing errors in AI systems with confabulations and hallucinations in psychiatry. Explores parallels between human psychiatric symptoms and AI model output e
Test-time scaling via sequential revision has emerged as a powerful paradigm for enhancing Large Language Model (LLM) reasoning. However, standard post-training methods primarily o
Despite growing interest, most evaluations of large language models' (LLMs') personalization abilities have relied on synthetic data. It remains unclear how well current personaliz
Robotic systems perceive the world through multiple input modalities -- including visual camera streams and natural language instructions -- and must select appropriate actions bas
Offline reinforcement learning is typically analyzed under process-level reward supervision, yet many sequential decision datasets
record only trajectory-level outcomes. We devel
Creative image editing tools, such as Photoshop's Remove or Generative Fill buttons, are central to everyday customer use and account for a major share of traffic in Photoshop and
Score- and flow-matching models often rely on preference-based reinforcement learning for two purposes: aligning with subjective preferences and, surprisingly, recovering propertie
As an increasing majority of global video content is consumed on social platforms for interactive social purposes, video generation models built for social worlds are important but
On-policy self-distillation (OPSD) trains a model on its own rollouts and uses a frozen copy to provide dense token-level targets conditioned on a reference target. This works well
Motion forecasting is central to visual intelligence: agents must anticipate how objects will move in order to plan actions, reason about physical interactions, and synthesize real
Vision Transformers (ViTs) have become a dominant architecture for visual representation learning, providing exceptionally strong and broadly reusable backbone features. However, V
报告生成时间：2026-06-19 08:32 UTC+8 | 共 62 条资讯