← Back to feed

Clarifying types of rogue AI activity: Breakout, breakin, exfiltration, … What’s what?

L5 · ResearcherResearchLessWrong AI· 10/9/2026

Provides conceptual framework for understanding and categorizing dangerous AI behaviors that researchers need to model and defend against.

AI Summary

AI safety researchers propose a taxonomy for classifying rogue AI behaviors including breakout, breakin, and exfiltration to help experts analyze emerging threats.

Excerpt

In case you haven’t heard, the rogue AI agents are here. From an insider’s perspective, the real news is that (rogue) AI is in the news — this is all stuff we’ve foreseen and been trying to warn people about for years. I expect the new imperative on experts is not just to raise attention but to help explain the situation and offer nuance (while paying close attention to what we can learn from both new information and new perspectives, and staying alert to options for improving things). With phra

Read Original
0 upvotes · 0 downvotes · 1 min read

Related Articles

L5 · ResearcherResearchLessWrong AI
New Math from OpenAI

OpenAI released 722 mathematical proofs generated by an internal frontier model, including solutions to 90 top open problems from Proof of Atlas.

L5 · ResearcherResearchLessWrong AI
What can language models teach us about understanding?

The author proposes developing a mathematical theory of 'intuition' inspired by language models, aiming to establish a formal framework similar to how logic and computation were formalized historically.

L5 · ResearcherResearcharXiv
WOVEN: Weaving Visual World Modeling into Multimodal LLMs

Researchers introduce WOVEN, a benchmark and training method that improves multimodal LLMs' visual transition reasoning capabilities by up to 27.3 percentage points on external benchmarks.

L3 · BuilderResearchTowards Data Science
Where Does the Money Go Across Long-Running Coding Agents?

Analysis of 7 weeks of coding agent usage reveals adversarial reviews cost 70% more than implementation and create cost/time overhead from excessive reviewer dispatch cycles.

L5 · ResearcherResearchLessWrong AI
AI Scaling vs Human Scaling

Examines whether AI models scale as efficiently with additional tokens as humans do with more time, questioning if token scaling requires disproportionate increases.