The different levels of how Claude thinks Anthropic 2026-09-25 | The different levels of how Claude thinks
Building Anthropic | A conversation with our co-founders Anthropic 2024-12-20 | The co-founders of Anthropic discuss the past, present, and future of Anthropic. From left to right: Chris Olah, Jack Clark, Daniela Amodei, Sam McCandlish, Tom Brown, Dario Amodei, and Jared Kaplan. Links and further reading:Anthropic's Responsible Scaling Policy (RSP): anthropic.com/news/announcing-our-updated-responsible-scaling-policy Machines of Loving Grace: darioamodei.com/machines-of-loving-graceWork with us: anthropic.com/careersClaude: claude.com00:00 Why work on AI?02:08 Scaling breakthroughs03:30 Early days of AI10:57 Sentiment shifting18:30 The Responsible Scaling Policy30:42 Founding story32:45 Building a culture of trust39:08 Racing to the top43:43 Looking to the future
Alignment faking in large language models Anthropic 2024-12-18 | Most of us have encountered situations where someone appears to share our views or values, but is in fact only pretending to do so—a behavior that we might call “alignment faking”.Could AI models also display alignment faking? Ryan Greenblatt, Monte MacDiarmid, Benjamin Wright and Evan Hubinger discuss a new paper from Anthropic, in collaboration with Redwood Research, that provides the first empirical example of a large language model engaging in alignment faking without having been explicitly—or even, we argue, implicitly—trained or instructed to do so.Learn more: anthropic.com/research/alignment-faking0:00 Introduction 0:47 Core setup and key findings of the paper6:14 Understanding alignment faking through real-world analogies9:37 Why alignment faking is concerning14:57 Examples of of model outputs21:39 Situational awareness and synthetic documents28:00 Detecting and measuring alignment faking38:09 Model training results47:28 Potential reasons for model behavior53:38 Frameworks for contextualizing model behavior1:04:30 Research in the context of current model capabilities1:09:26 Evaluations for bad behavior1:14:22 Limitations of the research1:20:54 Surprises and takeaways from results1:24:46 Future directions
What do people use AI models for? Anthropic 2024-12-12 | Despite the rapidly-growing popularity of large language models, until now we’ve had little insight into exactly how they’re being used. Claude insights and observations, or “Clio”, is our attempt to answer this question. Clio is an automated analysis tool that enables privacy-preserving analysis of real-world language model use. In this video, members of Anthropic's Societal Impacts team—Deep Ganguli, Esin Durmus, Miles McCain and Alex Tamkin—discuss the development of Clio, what they found, and the importance of this research.Read more: anthropic.com/research/clio0:00 Introduction02:57 What is Clio? 06:53 How we built Clio09:12 Privacy and ethical considerations13:32 How does Clio work?17:14 Findings and surprises22:24 How do we know Clio works?24:28 Trust and safety applications33:40 Real-world applications39:26 Why is this important?43:09 Future directions
Asana supercharges work management with Claude Anthropic 2024-11-07 | Asana, a leading enterprise work management platform, powers their Asana AI offering with Claude for its 150,000+ global customers. Their innovation continues to prove that AI is no longer just a tool, it’s a teammate that can advise teams on where to focus, action work and workflows, and adapt to how an organization works.Find out more: anthropic.com/customers/asana
Claude | Computer use for coding Anthropic 2024-10-22 | With the upgraded Claude 3.5 Sonnet, we’re introducing a new capability in beta: computer use. Developers can now direct Claude to use computers the way people do—by looking at a screen, moving a cursor, clicking, and typing text.While groundbreaking, computer use is still experimental—at times cumbersome and error-prone. We're releasing computer use early for feedback from developers.In this demo, Claude creates a themed website—generating code, launching a server, and fixing its own mistakes.Claude is generating all the computer actions shown here. This demonstration was recorded in a controlled environment, with some supporting infrastructure simplified to highlight the core capabilities.Read more about Claude and computer use: anthropic.com/news/3-5-models-and-computer-use
Claude | Computer use for orchestrating tasks Anthropic 2024-10-22 | With the upgraded Claude 3.5 Sonnet, we’re introducing a new capability in beta: computer use. Developers can now direct Claude to use computers the way people do—by looking at a screen, moving a cursor, clicking, and typing text.While groundbreaking, computer use is still experimental—at times cumbersome and error-prone. We're releasing computer use early for feedback from developers.In this demo, Claude orchestrates a multi-step task by searching the web, using native applications, and creating a plan with the resulting information.Claude is generating all the computer actions shown here. This demonstration was recorded in a controlled environment, with some supporting infrastructure simplified to highlight the core capabilities.Read more about Claude and computer use: anthropic.com/news/3-5-models-and-computer-use
Claude | Computer use for automating operations Anthropic 2024-10-22 | With the upgraded Claude 3.5 Sonnet, we’re introducing a new capability in beta: computer use. Developers can now direct Claude to use computers the way people do—by looking at a screen, moving a cursor, clicking, and typing text.At this stage, it is still experimental—at times cumbersome and error-prone. We're releasing computer use early for feedback from developers, and expect the capability to improve rapidly over time.In this demo, Claude searches through different tabs, gathers the requested information, and fills out a form—a task that could be scaled across many domains.Claude is generating all the computer actions shown here. This demonstration was recorded in a controlled environment, with some supporting infrastructure simplified to highlight the core capabilities.Read more about Claude and computer use: anthropic.com/news/3-5-models-and-computer-use
European Parliament expands access to their archives with Claude Anthropic 2024-10-21 | The European Parliament is revolutionizing access to its vast historical archives by implementing 'Ask the EP Archives', also known as Archibot, an AI-powered tool built with Claude in Amazon Bedrock that allows users to instantly search, retrieve, and understand decades of legislative documents and parliamentary proceedings in multiple languages.Learn more: anthropic.com/customers/european-parliament
AI, policy, and the weird sci-fi future with Anthropic’s Jack Clark Anthropic 2024-09-30 | A conversation with Jack Clark, Co-founder and Head of Policy at Anthropic, about the state of AI technology, how policymakers should think about AI, and the weird AI future that’s already beginning to emerge. Learn more about Anthropic: anthropic.com Sign up to Jack’s newsletter, ImportAI: importai.substack.com0:00 Introduction1:46 How governments are thinking about AI 5:06 Potential economic and social impacts of AI8:15 AI models as creative mirrors10:35 The “rogue state” theory of AI15:08 Challenges of machine time vs. human time20:12 Communities of AI explorers22:50 Fiction becomes reality27:15 AI’s good and bad potential30:22 AI policy in 202533:34 Key messages to policymakers35:11 Recommendations
Tips For Better Results With AI Anthropic 2024-09-16 | Anthropic's prompt engineering experts share tips to improve your AI results.
AI Prompts That Drive Growth Anthropic 2024-09-16 | Anthropic's David Hershey explains the difference between enterprise prompting and regular chat prompting.
AIs Greatest Challenge: You? Anthropic 2024-09-13 | Anthropic's David Hershey explains one of the big reasons why you're not getting what you want out of an AI.
Ancient Wisdom, Modern AI? Anthropic 2024-09-13 | Anthropic's Amanda Askell explains the unexpected parallels between philosophy and AI prompt engineering.
AI Prompt Engineering 101: Explained Anthropic 2024-09-13 | Anthropic's Zack Witten explains what prompt engineering is.
AI prompt engineering: A deep dive Anthropic 2024-09-05 | Some of Anthropic's prompt engineering experts—Amanda Askell (Alignment Finetuning), Alex Albert (Developer Relations), David Hershey (Applied AI), and Zack Witten (Prompt Engineering)—reflect on how prompt engineering has evolved, practical tips, and thoughts on how prompting might change as AI capabilities grow. Timestamps:0:00 Introduction2:05 Defining prompt engineering6:34 What makes a good prompt engineer12:17 Refining prompts24:27 Honesty, personas and metaphors in prompts37:12 Model reasoning45:18 Enterprise vs research vs general chat prompts50:52 Tips to improve prompting skills53:56 Jailbreaking56:51 Evolution of prompt engineering1:04:34 Future of prompt engineeringLearn more about Anthropic: anthropic.comAnthropic prompt engineering docs: docs.anthropic.com/en/docs/build-with-claude/prompt-engineering/overview
Wedia advances digital asset management with Claude Anthropic 2024-08-28 | Wedia offers a digital asset management platform that helps enterprises store, share, and organize their visual content. Anthropic’s Claude in Amazon Bedrock powers Wedia’s platform, supporting its team’s mission to help companies create secure, personalized customer experiences at scale.Learn more: anthropic.com/customers/wedia-group
How we built Artifacts with Claude Anthropic 2024-08-27 | Today, we’re making Artifacts available for all Claude.ai users. In addition, you can now create and view Artifacts on our iOS and Android apps. Artifacts turn conversations with Claude into a more creative and collaborative experience.With Artifacts, you have a dedicated window to instantly see, iterate, and build on the work you create with Claude. Since launching as a feature preview in June, users have created tens of millions of Artifacts.Here’s how it all started.Learn more: anthropic.com/news/artifactsTry Artifacts with Claude: claude.ai
Shareable Artifacts in Claude Anthropic 2024-07-10 | Artifacts made with Claude can now be published and shared! You can also remix Artifacts shared by others. This is an experimental feature. Remix an example Artifact to get started: https://claude.site/artifacts/c55cf857-d456-4520-8ee2-206697dfa2a3.You can ask Claude to generate docs, code, mermaid diagrams, vector graphics, or even simple games. Artifacts appear next to your chat, letting you see, iterate, and build on your creations in real-time.
Evaluate prompts in the Anthropic Console Anthropic 2024-07-09 | Generate, test, and evaluate your prompts in the Anthropic Console—including the ability to generate automatic test cases and compare outputs—so Claude generates the very best responses for your needs. Get started on console.anthropic.com or read more on anthropic.com/news/evaluate-prompts.
Anthropic at 42Paris Anthropic 2024-06-26 | We recently ran a workshop at 42Paris to speak to the community about using Claude. Try Claude: claude.aiFind out more about Anthropic: anthropic.comFind out more about 42Paris: https://42.fr
Shareable Projects in Claude Anthropic 2024-06-25 | You can now organize chats with Claude into shareable Projects.Projects allow you to ground Claude's outputs in your internal knowledge—be it style guides, codebases, interview transcripts, or past work. Each project includes a 200K context window, the equivalent of a 500-page book.In this demo, we create a project for new team members, adding docs on working practices, team charters, all-hands notes, and more. We then ask Claude to make an org chart diagram, save it, before we email the project link to a new teammate. They join and start chatting with Claude, armed with all of the knowledge we've added.Learn more: anthropic.com/news/projects
Claude 3.5 Sonnet for agentic coding Anthropic 2024-06-20 | Claude 3.5 Sonnet sets new industry benchmarks for coding proficiency.With Claude, you can go you from an incomplete implementation to a fully functioning one, including unit tests, with minimal input from you.Find out more: anthropic.com/news/claude-3-5-sonnet Try Claude: claude.ai
Claude 3.5 Sonnet as a writing partner Anthropic 2024-06-20 | Claude 3.5 Sonnet sets new industry benchmarks for graduate-level reasoning (GPQA), undergraduate-level knowledge (MMLU), and coding proficiency (HumanEval). It shows marked improvement in grasping nuance, humor, and complex instructions, all while writing with a natural tone.Find out more: anthropic.com/news/claude-3-5-sonnet Try Claude: claude.ai
Claude 3.5 Sonnet for vision Anthropic 2024-06-20 | Claude 3.5 Sonnet is our strongest vision model.Improvements are most noticeable in tasks requiring visual reasoning, like interpreting charts, graphs, or transcribing text from imperfect images.Find out more: anthropic.com/news/claude-3-5-sonnet Try Claude: claude.ai
Claude 3.5 Sonnet for sparking creativity Anthropic 2024-06-20 | Claude 3.5 Sonnet can be used with Artifacts—a new feature that expands how users can interact with Claude. You can ask Claude to generate docs, code, mermaid diagrams, vector graphics, or even simple games. Artifacts appear next to your chat, letting you see, iterate, and build on your creations in real-time.Find out more: anthropic.com/news/claude-3-5-sonnet Try Claude: claude.ai
Scaling interpretability Anthropic 2024-06-13 | Science and engineering are inseparable. Our researchers reflect on the close relationship between scientific and engineering progress, and discuss the technical challenges they encountered in scaling our interpretability research to much larger AI models.Read more: anthropic.com/research/engineering-challenges-interpretability
What should an AIs personality be? Anthropic 2024-06-08 | How do you imbue character in an AI assistant? What does that even mean? And why would you do it in the first place? In this conversation, Stuart Ritchie (Research Communications at Anthropic) speaks to Amanda Askell (Alignment Finetuning Researcher at Anthropic) about the ins and outs of “character training” for Claude, Anthropic’s AI model. 00:00 — Introduction01:41 — The importance of an AI’s character03:32 — Training an AI model05:44 — System prompts07:24 — Impacts of system prompts11:20 — Character vs personality16:39 — Training good moral character19:10 — Claude's trait: charitability25:08 — Claude's trait: honesty 28:21 — Deciding on Claude’s personality31:11 — Self awareness in AI34:02 — Kindness towards AI37:25 — ConclusionLearn more about Claude’s character: anthropic.com/research/claude-characterLearn more about research at Anthropic: anthropic.com/research Try out Claude: claude.ai
What is interpretability? Anthropic 2024-06-03 | A surprising fact about modern large language models is that nobody really knows how they work internally. At Anthropic, the Interpretability team strives to change that — to understand these models to better plan for a future of safe AI.Find out more: anthropic.com/research
Claude is now available in Europe Anthropic 2024-05-15 | Claude is now available in Europe.Try it out: http://claude.aiDownload the iOS app: apps.apple.com/us/app/claude/id6473753684Find out more: anthropic.com/news/claude-europe
Introducing the Claude iOS app Anthropic 2024-05-03 | The Claude iOS app has arrived. The power of frontier intelligence is now in your back pocket. Chat with Claude whenever, and wherever, inspiration strikes. Claude on iOS syncs with your chat history, lets you ask questions about photos, and handles tasks on the go — powered by our industry-leading Claude 3 models.Download now on the App Store: apps.apple.com/app/id6473753684
Coming soon to the Team plan on Claude.ai Anthropic 2024-05-01 | Our new Team plan is designed for ambitious teams that want to collaborate with Claude on deep work, expert research, and tasks of every size. Learn more: anthropic.com/claude
Tool use with the Claude 3 model family Anthropic 2024-04-04 | Tool use, which enables the Claude 3 model family to interact with external tools, is now available in beta to all customers in the Anthropic Messages API.Tool use, or function calling, is a frontier AI capability that allows Claude to reason, plan and execute a set of actions by generating structured outputs via API calls.Read our developer documentation: docs.anthropic.com/claude/docs/tool-useTry out tool use: anthropic.com/apiLearn more: anthropic.com/claude
Claude 3 Haiku for fast document analysis Anthropic 2024-03-14 | Introducing Claude 3 Haiku, the fastest and most affordable AI model in its intelligence class.Learn more in our blog post: anthropic.com/news/claude-3-haikuTry Claude 3 now at claude.aiLearn more about Claude, and our model family anthropic.com/claude
Claude 3 Haiku for instant customer service Anthropic 2024-03-14 | Introducing Claude 3 Haiku, the fastest and most affordable AI model in its intelligence class.Learn more in our blog post: anthropic.com/news/claude-3-haikuVisit here to learn more about Claude, and our model family anthropic.com/claude
Claude 3 Haiku turns thousands of physical documents into structured data Anthropic 2024-03-04 | Introducing Claude 3, our next generation of AI models.The three state-of-the-art models—Claude 3 Opus, Claude 3 Sonnet, and Claude 3 Haiku—set new industry benchmarks across reasoning, math, coding, multilingual understanding, and vision.Learn more in our blog post: anthropic.com/news/claude-3-familyBuild with Claude: anthropic.com/api
Claude 3 Sonnet as a language learning partner Anthropic 2024-03-04 | Introducing Claude 3, our next generation of AI models.The three state-of-the-art models—Claude 3 Opus, Claude 3 Sonnet, and Claude 3 Haiku—set new industry benchmarks across reasoning, math, coding, multilingual understanding, and vision.Try this prompt for yourselves at claude.aiLearn more in our blog post: anthropic.com/news/claude-3-family
Claude 3 Opus as an economic analyst Anthropic 2024-03-04 | Introducing Claude 3, our next generation of AI models.The three state-of-the-art models—Claude 3 Opus, Claude 3 Sonnet, and Claude 3 Haiku—set new industry benchmarks across reasoning, math, coding, multilingual understanding, and vision.Learn more in our blog post: anthropic.com/news/claude-3-family
Robin AI, powered by Claude Anthropic 2024-01-24 | Robin AI uses Claude to reduce the time it takes to analyze and sign legal contracts - making the process up to 10 times faster than before. Their team chose to partner with Anthropic for its reliable systems and trusted outputs, as well as the security of its partnership with Amazon Bedrock.Try Robin AI here: robinai.comTalk to Claude at: claude.aianthropic.com
Behind the prompt: Prompting tips for Claude.ai Anthropic 2023-08-25 | Alex, a prompt engineer at Anthropic shares a little about his journey into prompting, and his 5 best tips for using Claude.ai. docs.anthropic.comclaude.aiRead the research paper: Red Teaming Language Models toReduce Harms — anthropic.com/index/red-teaming-language-models-to-reduce-harms-methods-scaling-behaviors-and-lessons-learned
Coding with Claude Anthropic 2023-08-23 | Add interactive data to a static map with Claude 2's coding capabilities.
Long inputs, multi-step output with Claude Anthropic 2023-08-23 | Summarize content and make a highlight table using Claude's 100K token long context window.
Inside our first Anthropic Hackathon, San Francisco Anthropic 2023-08-23 | Hear from AI enthusiasts and see what they #BuildwithClaude in this inside look at our first-ever Anthropic Hackathon in San Francisco.Talk to Claude at claude.aiRead our developer docs: docs.anthropic.comLearn about Constitutional AI anthropic.com/index/constitutional-ai-harmlessness-from-ai-feedback
Quick tips for Claude: Long context file uploads Anthropic 2023-08-23 | Upload long files to Claude for clear summaries.
