Aligning, Fast and Slow: A Control-Theoretic Oversight Framework for Long-Horizon Trajectories and Maintained Deployments
cs.AI
Submitted: 2026-11-05
Updated: 2026-11-05
Code: https://github.com/marshmallow-code/marshmallow
Terminology
Sources
- Position: A Three-Layer Probabilistic Assume-Guarantee Architecture Is Structurally Required for Safe LLM Agent Deployment
- Ctrl-Z: Controlling AI Agents via Resampling
- Memory-Induced Tool-Drift in LLM Agents
- Stateful Agent Backdoors: Constructing Cross-Session Attack Programs
- Self-Propagating Misalignment in LLM Agents, and Why Auditing or Disabling Memory Is Not Enough
- TRACE: Trajectory Risk-Aware Compression for Long-Horizon Agent Safety
- Enforcing Temporal Constraints for LLM Agents
- Evaluating Control Protocols for Untrusted AI Agents
- SHADE-Arena: Evaluating Sabotage and Monitoring in LLM Agents
- Hidden in Memory: Sleeper Memory Poisoning in LLM Agents
- Adaptive Attacks on Trusted Monitors Subvert AI Control Protocols
Related papers
- Multi-Objective Structured Pruning of LLMs for Latency and Model Size Optimization
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- Embeddings for Preferences, Not Semantics
- Can We Trust Item Response Theory for AI Evaluation?
- Diagnosing Task Insensitivity in Language Agents