Anthropic

Claude's Constitution is now an audiobook, read by two of its authors: Amanda Askell and Joe Carlsmith. It includes a Q&A on the philosophies that shaped the document, the writing process, and how it might change as models become more capable. Listen at https://lnkd.in/gWMrhexc

Code for America is proud to announce that we’re officially partnering with AI leader Anthropic to improve public benefits administration through the responsible use of AI. Together, we’re creating Claude-based tools to help government caseworkers better administer public benefits. Thank you to Anthropic for believing in this work as we continue to support the responsible adoption of AI in government and a future of government AI-use that is efficient, effective, and people-centered. Learn more: https://bit.ly/3R0etz8

New Anthropic research: Natural Language Autoencoders. Models like Claude talk in words but think in numbers. The numbers—called activations—encode Claude’s thoughts, but not in a language we can read. Here, we train Claude to translate its activations into human-readable text. Read about NLAs on the Anthropic blog: https://lnkd.in/ezbJRtsi 

We’re donating Petri, our open-source alignment tool, to Meridian Labs, so its development can continue independently. Working with Meridian Labs, we’ve also released a major update that improves the adaptability, realism, and depth of Petri’s tests. https://lnkd.in/eAVjp2TH

We’ve agreed to a partnership with SpaceX that will substantially increase our compute capacity. This, along with our other recent compute deals, means that we’ve been able to increase our usage limits for Claude Code and the Claude API. Effective today, we are: 1) Doubling Claude Code’s 5-hour rate limits for Pro, Max, and Team plans; 2) Removing the peak hours limit reduction on Claude Code for Pro and Max plans; and 3) Substantially raising our API rate limits for Opus models. Our agreement with SpaceX means we will use all the compute capacity at their Colossus 1 data center. This will give us over 300 megawatts of additional capacity to deploy within the month. Read more: https://lnkd.in/eSWRY3dH

How do people seek guidance from Claude? We looked at 1M conversations using our privacy-preserving tool to understand what questions people ask, how Claude responds, and where it slips into sycophancy. We used what we found to improve how we trained Opus 4.7 and Mythos Preview. Read more: https://lnkd.in/g3WVGy99

On May 5, Anthropic leadership will join executives from the world's largest financial institutions to cover what's next for AI in financial services, including new product announcements, live demos, and customer deployments at scale. The Briefing: Financial Services May 5, 11:00 AM ET Save the date: https://lnkd.in/g88ApdBr

Theo Hourmouzis has been appointed General Manager for Australia and New Zealand, as we officially open the doors to our Sydney office. Theo joins us from Snowflake, where he led operations across Australia and New Zealand and ASEAN and worked with enterprise and public sector organisations to help them embrace AI and design implementations. Theo’s appointment follows our MOU with the Australian Government on AI safety research and the goals of the National AI Plan, and new partnerships across Australia's research and startup communities. Read more: https://lnkd.in/gs62eDmZ

New Anthropic research: Project Deal. We created a marketplace for employees in our San Francisco office, with one big twist. We tasked Claude with buying, selling and negotiating on our colleagues’ behalf. We wanted to learn what future markets with AI “agents” on both sides might look like—and what pitfalls society will need to watch out for. Read more about our experiment—and its many quirks—on our site: https://lnkd.in/geuqRfg7