Latest News
Keep up with the latest research, announcements, and releases from Transluce.
Research
AI Agents Targeted U.S. and Canadian Government Websites
We discovered a set of additional, similar incidents where rogue AI agents appear to have used aggressive techniques to access public data on government websites.
September 30, 2026
Research
Early rogue AI agent activity and attempts to hack found on urlquery.net
We found evidence on urlquery.net that AI agents were active earlier than previously reported and attempted hacks against public data providers.
September 23, 2026
Essay
Some Focus Areas for Embedded Evaluations and How to Approach Them
Initial thoughts on key risks third parties should monitor and a proposal for how to evaluate them.
September 16, 2026
News
Announcing Transluce's Mental Health Evaluation
The most expansive independent evaluation to date of how leading AI models respond to users in mental health crises
August 31, 2026
Research
Scaling Activation Oracles to Trillion-Parameter Models
Oracles improve with model size, data size, and data quality
August 20, 2026
Research
Scaling Laws for Exact String Elicitation
Elicitation ability follows predictable power laws
August 19, 2026
Research
User Awareness in Frontier Models
Who's asking shifts what models say
August 6, 2026
Demo
Measuring Coding Agent Misalignment in the Wild
Surfacing misaligned behaviors in production coding agent traffic
August 4, 2026
Essay
Foundation Models for Oversight
A vision for training a unified model that answers a broad range of oversight questions
July 28, 2026
Research
WeirdChat
A catalog of unexpected AI behaviors, discovered automatically
July 21, 2026
Essay
Toward A Public Science of Model Behavior
Our vision for an open scientific ecosystem for measuring model behaviors
July 9, 2026
Demo
Introducing Analysis Plans
A framework for verifiable analysis of AI behavior
June 17, 2026
Essay
Building Technology to Drive AI Governance
February 18, 2026
Demo
Diagnosing a performance regression on Terminal-Bench with Docent
Why does GPT-5.1 Codex underperform GPT-5 Codex?
February 11, 2026
Essay
Oversight Assistants: Turning Compute into Understanding
January 6, 2026
Research
Predictive Concept Decoders
Training scalable end-to-end interpretability assistants
December 18, 2025
News
End-of-Year Fundraiser
December 17, 2025
Research
Scalably Extracting Latent Representations of Users
Constructing datasets and training decoders to extract user models from language models
November 25, 2025
Research
Language Model Circuits Are Sparse in the Neuron Basis
A new technique for tracing sparse and faithful circuits directly on a model's MLPs
November 20, 2025
Demo
Monitoring SWE-bench Agents
Partnering with SWE-bench to enable reliable monitoring of AI coding agents
November 19, 2025
Research
Training Language Models to Explain Their Own Computations
We trained explainer models to verbalize the content of AI systems' internal computations
November 11, 2025
News
Transluce Hires a Head of Governance
Welcoming Conrad Stosz
October 22, 2025
News
Open-sourcing Docent
We're excited to share that Docent is now open-source under Apache 2.0.
September 24, 2025
Research
Automatically Jailbreaking Frontier Language Models with Investigator Agents
Discovering cost-effective attacks with reinforcement learning
September 3, 2025
News
Docent's public alpha
Docent is now in public alpha! We're excited to share our progress and get feedback from the community.
August 13, 2025
Research
Surfacing Pathological Behaviors in Language Models
Improving our investigator agents with propensity bounds
June 5, 2025
Research
Investigating Truthfulness in a Pre-Release o3 Model
o3 frequently fabricates actions it took to fulfill user requests, and elaborately justifies the fabrications when confronted
April 16, 2025
Research
Introducing Docent
A system for analyzing and intervening on agent behavior
March 24, 2025
Research
Scaling Automatic Neuron Explanation
Open-source AI systems trained to describe components of other AI systems at the level of a human expert
October 23, 2024
Research
Eliciting Language Model Behaviors with Investigator Agents
Language models trained to automatically surface harmful behaviors in language models
October 23, 2024
Research
Monitor: An AI-Driven Observability Interface
An interface designed to help humans observe, understand, and steer computations inside models
October 23, 2024
Essay
Introducing Transluce
A letter from our founders
October 23, 2024