A look at Anthropic's Clio, an internal AI tool to identify new threats, disrupt coordinated abuse of the company's systems, and generate Claude usage insights
PLUS: Exclusive data on how people are using Anthropic's chatbot — The company didn't know it yet, but Anthropic had a spam problem.
Platformer Casey Newton
Related Coverage
- Clio: A system for privacy-preserving insights into real-world AI use Anthropic
- Clio: A system for privacy-preserving insights into real-world AI use. … Simon Willison's Weblog
- Clio: Privacy-Preserving Insights into Real-World AI Use Anthropic
- What do people use AI models for? Anthropic on YouTube
- Anthropic Unveils its Clio Framework For Claude Usage Tracking and Threat Detection WinBuzzer · Markus Kasanmascheff
- Anthropic Introduces Clio: A New AI System that Automatically Identifies Trends in Claude Usage Across the World MarkTechPost · Aswin Ak
- Anthropic outlines most popular Claude use cases Constellation Research · Larry Dignan
- I saw some of this previewed during new hire orientation and it's v. slick. — I don't pretend to understand how it works to cluster while preserving privacy but I like it. … Todd Underwood
- Clio: A system for privacy-preserving insights into real-world AI use Hacker News
Discussion
-
@renee.diresta
Renee DiResta
on threads
Love seeing brilliant SIO alums in good stories! “Clio is an internal tool at Anthropic that uses machine learning to identify previously unknown threats, disrupt coordinated attempts to abuse the company's systems, and generate insights about how Claude and the company's artifi…
-
@crumbler
Casey Newton
on threads
Clio is Anthropic's new system for identifying AI risks that it hadn't thought to look for — what it calls the unknown unknowns. I talked with team that built it and share for the first time the top three ways people use Claude https://www.platformer.news/ ... [image]
-
@anthropicai
@anthropicai
on x
Clio can help us spot abuses of the model, and also find false positives (where benign uses are inadvertently flagged as dangerous). [image]
-
@simonw
Simon Willison
on x
Fascinating insight from Anthropic here concerning how Claude is actually being used, via their Clio system that attempts to preserve end-user privacy while still exposing clusters of usage patterns (using a clever design based on embeddings) My notes: https://simonwillison.net/.…
-
@anthropicai
@anthropicai
on x
New Anthropic research: How are people using AI systems in the real world? We present a new system, Clio, that automatically identifies trends in Claude usage across the world. [image]
-
@anthropicai
@anthropicai
on x
Clio works by taking real-world conversations, summarizing them while removing identifying information, and clustering these summaries to reveal aggregate insights. This process happens in a secure environment—only the final, high-level insights are visible to human analysts. [im…