LLM Digest

Story

arxiv_cs_ai · Nov 12, 2026 · paper

Source brief

S1-Evo: Experience-Guided Evidence Competition Attacks on Typed System-1 Decision Models

arxiv.orgNov 12, 2026
original source linked

In brief

Typed System-1 (S1) decision models are increasingly deployed in LLM agents for routing, moderation, and authorization to cut inference cost, yet untrusted context can still manipulate typed outputs and alter executio...

Feed lens

agent

Read the original at arxiv.org →Open in live feed Read that day’s brief

S1-Evo: Experience-Guided Evidence Competition Attacks on Typed System-1 Decision Models

Earlier in this thread 4 items

Measure Before You Train: Typed Decision Models vs. a Classical Baseline vs. a Generative LLM for Routing a Personal Multilingual Inbox

One Word Opens the Gate: The Option-Channel Attack on Typed Decision Models as Agent Guardrails

Presentation Sensitivity in Typed Decision Models for Subjective Text Evaluation

VX-One: Typed, Calibrated Decision Heads for Real-Time Perimeter Triage - An empirical study of unknown-attack honesty, cascade gray zones, and calibration under drift