From hard refusals to safe-completions: toward output-centric safety training - 惯性聚合

推荐订阅源

Recent Commits to openclaw:main

OSCHINA 社区最新新闻

博客园 - 【当耐特】

宝玉的分享

让小产品的独立变现更简单 - ezindie.com

Last Week in AI

人人都是产品经理

博客园_首页

Tailwind CSS Blog

Proofpoint News Feed

Darknet – Hacking Tools, Hacker News & Cyber Security

Vulnerabilities – Threatpost

阮一峰的网络日志

CERT Recently Published Vulnerability Notes

Attack and Defense Labs

Schneier on Security

Security Archives - TechRepublic

Lohrmann on Cybersecurity

Visual Studio Blog

WordPress大学

The Register - Security

Netflix TechBlog - Medium

Hugging Face - Blog

博客园 - 叶小钗

Full Disclosure

大猫的无限游戏

SegmentFault 最新的问题

Cyber Attacks, Cyber Crime and Cyber Security

Google Online Security Blog

Recorded Future

Cyber Security Advisories - MS-ISAC

Hacker News - Newest: "LLM"

LINUX DO - 最新话题

Google DeepMind News

Privacy International News Feed

Fortinet All Blogs

Hackread – Cybersecurity News, Data Breaches, AI and More

OpenAI News

Using custom GPTs ChatGPT for customer success teams Applications of AI at OpenAI Research with ChatGPT Analyzing data with ChatGPT Financial services Responsible and safe use of AI Writing with ChatGPT ChatGPT for research Creating images with ChatGPT Personalizing ChatGPT ChatGPT for finance teams Getting started with ChatGPT Working with files in ChatGPT ChatGPT for sales teams Prompting fundamentals ChatGPT for managers Using projects in ChatGPT ChatGPT for marketing teams Brainstorming with ChatGPT AI fundamentals ChatGPT for operations teams Healthcare Our response to the Axios developer tool compromise Using skills CyberAgent moves faster with ChatGPT Enterprise and Codex The next phase of enterprise AI Introducing the Child Safety Blueprint Introducing the OpenAI Safety Fellowship Industrial policy for the Intelligence Age OpenAI acquires TBPN Codex now offers more flexible pricing for teams Gradient Labs gives every bank customer an AI account manager OpenAI raises $122 billion to accelerate the next phase of AI Helping disaster response teams turn AI into action across Asia STADLER reshapes knowledge work at a 230-year-old company Inside our approach to the Model Spec Introducing the OpenAI Safety Bug Bounty program Helping developers build safer AI experiences for teens Update on the OpenAI Foundation Powering Product Discovery in ChatGPT Creating with Sora Safely How we monitor internal coding agents for misalignment OpenAI to acquire Astral Introducing GPT-5.4 mini and nano OpenAI Japan announces Japan Teen Safety Blueprint to put teen safety first Equipping workers with insights about compensation Why Codex Security Doesn’t Include a SAST Report Designing AI agents to resist prompt injection From model to agent: Equipping the Responses API with a computer environment Rakuten fixes issues twice as fast with Codex Wayfair boosts catalog accuracy and support speed with OpenAI Improving instruction hierarchy in frontier LLMs New ways to learn math and science in ChatGPT OpenAI to acquire Promptfoo Codex Security: now in research preview How Descript engineers multilingual video dubbing at scale How Balyasny Asset Management built an AI research engine Reasoning models struggle to control their chains of thought, and that’s good Introducing GPT-5.4 GPT-5.4 Thinking System Card Ensuring AI use in education leads to opportunity VfL Wolfsburg turns ChatGPT into a club-wide capability Introducing ChatGPT for Excel and new financial data integrations Introducing the Adoption news channel The five AI value models driving business reinvention Extending single-minus amplitudes to gravitons Understanding AI and learning outcomes How Axios uses AI to help deliver high-impact local journalism GPT-5.3 Instant System Card GPT-5.3 Instant: Smoother, more useful everyday conversations Our agreement with the Department of War OpenAI and Amazon announce strategic partnership Joint Statement from OpenAI and Microsoft Introducing the Stateful Runtime Environment for Agents in Amazon Bedrock Scaling AI for everyone An update on our mental health-related work Pacific Northwest National Laboratory and OpenAI partner to accelerate federal permitting OpenAI Codex and Figma launch seamless code-to-design experience Advancing independent research on AI alignment Introducing OpenAI for India Introducing EVMbench GPT-5.2 derives a new result in theoretical physics Introducing Lockdown Mode and Elevated Risk labels in ChatGPT Beyond rate limits: scaling access to Codex and Sora Scaling social science research Introducing GPT-5.3-Codex-Spark Harness engineering: leveraging Codex in an agent-first world Bringing ChatGPT to GenAI.mil Testing ads in ChatGPT Making AI work for everyone, everywhere OpenAI and NORAD team up to bring new magic to “NORAD Tracks Santa” Accenture and OpenAI accelerate enterprise AI success OpenAI takes an ownership stake in Thrive Holdings to accelerate enterprise AI adoption What to know about a recent Mixpanel security incident Expanding data residency access to business customers worldwide Inside JetBrains—the company reshaping how the world writes code Introducing shopping research in ChatGPT How GPT-5 helped mathematician Ernest Ryu solve a 40-year-old open problem OpenAI and Foxconn collaborate to strengthen U.S. manufacturing across the AI supply chain

From hard refusals to safe-completions: toward output-centric safety training

2025-08-07 · via OpenAI News

此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。原文来自 — 版权归原作者所有。