TEPA: Revoking Stale Memories for Conflict-Robust Language Agents

Long-term memory enables language agents to reuse past facts, preferences, and task experience. Persistence also creates a central falsifiability problem: when the world changes, stale memories can remain retrievable and pollute the prompt. We characterize this failure mode as memory pollution: degradation caused by active memories that newer conflicting evidence has superseded. We introduce…

arxiv.org ·2026-08-07

We haven't written up this one. arXiv cs.AI has the full story — the link below goes straight to it.

Read the original at arxiv.org →

More in AI

A Picture is Worth a Thousand Tokens: How Vision Language Models Cut AI Energy Costs While Improving Accuracy

LLM inference accounts for over 90% of AI operational energy, scaling directly with input token count---a critical inefficiency for telecom network analytics and numerical time-series data analysis…

2026-08-07 ·arxiv.org

CoBa: Cost-Effective Test-Time Scaling via Compute-Balanced Routing

Test-time scaling is often implemented by spending more compute along one axis: sampling more solutions, extending a chain of thought, or applying a stronger evaluator.

2026-08-07 ·arxiv.org

Stanford is running 37,000 AI agents as a virtual biotech — and one of its drug designs got independently confirmed by Merck

For developers, the operating assumption has been one engineer, one agent — the model Claude Code and similar tools. At VB Transform 2026 , James Zou, associate professor of biomedical data science at…

2026-08-07 ·venturebeat.com

TEPA: Revoking Stale Memories for Conflict-Robust Language Agents

More in AI

A Picture is Worth a Thousand Tokens: How Vision Language Models Cut AI Energy Costs While Improving Accuracy

CoBa: Cost-Effective Test-Time Scaling via Compute-Balanced Routing

Stanford is running 37,000 AI agents as a virtual biotech — and one of its drug designs got independently confirmed by Merck

Diffusion LLMs as Targets and Adversaries: Mechanistic Safety Exploits

SABRE: Scalable and Automated Benchmarking of VLMs under Stress

Post-Grokking Collapse at the Representation-Readout Interface in Muon-Trained Transformers