VerificAgent: Domain-Specific Memory Verification for Scalable Oversight of Aligned Computer-Use Agents
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2025
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866912526643494912 |
|---|---|
| author | Nguyen, Thong Q. Desai, Shubhang Anwar, Raja Hasnain Shaik, Firoz Suryanarayanan, Vishwas Chowdhary, Vishal |
| author_facet | Nguyen, Thong Q. Desai, Shubhang Anwar, Raja Hasnain Shaik, Firoz Suryanarayanan, Vishwas Chowdhary, Vishal |
| contents | Continual memory augmentation lets computer-using agents (CUAs) learn from prior interactions, but unvetted memories can encode domain-inappropriate or unsafe heuristics--spurious rules that drift from user intent and safety constraints. We introduce VerificAgent, a scalable oversight framework that treats persistent memory as an explicit alignment surface. VerificAgent combines (1) an expert-curated seed of domain knowledge, (2) iterative, trajectory-based memory growth during training, and (3) a post-hoc human fact-checking pass to sanitize accumulated memories before deployment. Evaluated on OSWorld productivity tasks and additional adversarial stress tests, VerificAgent improves task reliability, reduces hallucination-induced failures, and preserves interpretable, auditable guidance--without additional model fine-tuning. By letting humans correct high-impact errors once, the verified memory acts as a frozen safety contract that future agent actions must satisfy. Our results suggest that domain-scoped, human-verified memory offers a scalable oversight mechanism for CUAs, complementing broader alignment strategies by limiting silent policy drift and anchoring agent behavior to the norms and safety constraints of the target domain. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2506_02539 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | VerificAgent: Domain-Specific Memory Verification for Scalable Oversight of Aligned Computer-Use Agents Nguyen, Thong Q. Desai, Shubhang Anwar, Raja Hasnain Shaik, Firoz Suryanarayanan, Vishwas Chowdhary, Vishal Machine Learning Continual memory augmentation lets computer-using agents (CUAs) learn from prior interactions, but unvetted memories can encode domain-inappropriate or unsafe heuristics--spurious rules that drift from user intent and safety constraints. We introduce VerificAgent, a scalable oversight framework that treats persistent memory as an explicit alignment surface. VerificAgent combines (1) an expert-curated seed of domain knowledge, (2) iterative, trajectory-based memory growth during training, and (3) a post-hoc human fact-checking pass to sanitize accumulated memories before deployment. Evaluated on OSWorld productivity tasks and additional adversarial stress tests, VerificAgent improves task reliability, reduces hallucination-induced failures, and preserves interpretable, auditable guidance--without additional model fine-tuning. By letting humans correct high-impact errors once, the verified memory acts as a frozen safety contract that future agent actions must satisfy. Our results suggest that domain-scoped, human-verified memory offers a scalable oversight mechanism for CUAs, complementing broader alignment strategies by limiting silent policy drift and anchoring agent behavior to the norms and safety constraints of the target domain. |
| title | VerificAgent: Domain-Specific Memory Verification for Scalable Oversight of Aligned Computer-Use Agents |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2506.02539 |