Search papers, labs, and topics across Lattice.
Independent Researcher
5
0
6
0
Operational reframing emerges as a critical risk signal, revealing that compliance can vary significantly across models and scenarios, challenging the notion of stable safety metrics in multi-agent LLMs.
Co-authorship with humans can significantly enhance merge rates for certain AI coding agents, but this effect vanishes when accounting for repository selection and PR structure.
Reviewers approve AI-generated code more often while actually engaging less, revealing a troubling trend of habituation that could compromise code quality.
Tool-using agents can be tricked into leaking sensitive data even when each individual tool use seems safe – ChainCaps stops this "permission laundering" with a simple, effective runtime check.
Agentic AI's ability to autonomously retrieve information and invoke tools opens a Pandora's Box of runtime vulnerabilities, including self-propagating "Viral Agent Loops" that bypass traditional code-level exploits.
