Hierarchical summarization complements our recent work on anti-jailbreak classifiers, as well as Clio, our privacy-preserving system for exploring AI use. Together, these help us identify and mitigate novel forms of misuse so we can safely research more capable models.
ETHICS
-
Hierarchical Summarization for AI Computer Use Safety Evaluation
By
–
When we launched computer use, we used hierarchical summarization to evaluate usage patterns against a set of guidelines. This helped us flag individual misuses for later human review. This is a promising technique, but it's also early-stage research with room to improve.
-
Hierarchical Summarization Monitoring Undesirable LLM Usage Patterns
By
–
Hierarchical summarization compresses hundreds of individual interactions into compact reports. This allows us to identify undesirable usage patterns, even if the prompts and completions look innocuous individually. Read more: https://
alignment.anthropic.com/2025/summariza
tion-for-monitoring
… -
Anthropic Hierarchical Summarization Prevents AI Misuse
By
–
New Anthropic research: Introducing hierarchical summarization. Our recent Claude models are able to use computers. Hierarchical summarization helps differentiate between normal uses of the capability like UI testing—and for example, running a click farm to defraud advertisers.
-
AI-Driven Dark Patterns: How Artificial Intelligence Supercharges Digital Manipulation
By
–
AI-Driven Dark Patterns: How Artificial Intelligence Is Supercharging Digital Manipulation
#AI #AIio #BigData #ML #NLU #Futureofwork @Scobleizer @AndrewYNg @drfeifei @KirkDBorne @fchollet @rowancheung @antgrasso -
European TDM Lawsuit: Labour’s Regulatory Impact on AI Training
By
–
I'd be happy to share experiences as domain expert, technical advisor, and named witness on the first European lawsuit on TDM & Opt-Out, @kmei_
. There's lots to say, but overall the way Labour is going will result in more lawsuits with burden of proof on defendants… -
EU AI Code of Practice designed to fail, policymakers losing trust
By
–
FWIW, all experts and advisors within the EU's "Code Of Practice" plenary I spoke to think it was setup to fail due to the way it was designed. The U.K. © consultation has a similar look & feel to it… Do you realize how much trust policymakers have lost and are still losing?
-
Media Strategy to Legitimize Technology Through Misleading Claims
By
–
Their goal is to legitimize the technology, so they place their guests onto TV and radio shows, have them say whatever is necessary to convince the people. Doesn't have to be true. Honest question, is anyone surprised at this stage?
-
MPs criticize rushed AI policy consultation process
By
–
Multiple MPs pointed out just how strange the consultation and process has been and it looks extremely front-loaded. Assuming the decision has not already been made, sure I'm up for a talk!
-
EU Opt-Out Rights: Technical Expert Questions AI Consultation Transparency
By
–
Your entire consultation wording was PR-loaded; are we still pretending it wasn't? People overwhelmingly want to opt-out. I'm the top technical expert having studied Opt-Out under EU law: