analiza aiqbench
This research addresses the critical challenge of "right to be forgotten" in persistent large language models, demonstrating that naive deletion fails due to semantic dependencies within conversation histories. The authors introduce the DeLLM framework, which utilizes a provenance graph to dynamically reconstruct context and effectively erase specific memories while preserving model utility. This work advances AI safety and privacy by establishing rigorous methods for managing memory retention and deletion in long-context systems.
Eternal Sunshine of the Spotless Mind: Systematically Erasing LLM's Memories
arXiv:2609.36414v1 Announce Type: new Abstract: We consider persistent LLMs that accumulate memories of their interactions with a user over time. Such LLMs maintain memories using external storage, which they can query to overcome the limitations of a fixed context window. Such systems have numerous practical applications, as they can draw on all past interactions when responding to user queries. In this paper, we ask whether LLMs can forget information shared with them upon a user's request. We find that current LLMs fail to delete such information---even when they claim to have forgotten it and even when operating with a limited context. To this end, we consider a new direction of study: Deletion of LLM Memories. We show that naively removing messages that match a user's deletion request is insufficient, since conversations naturally introduce message dependencies that cause information to persist. To correctly handle deletion requests, we propose the DeLLM framework. It dynamically constructs relevant context for each LLM query and maintains a provenance graph of messages to determine which ones must be removed during deletion. Our experiments show that DeLLM achieves a high deletion rate while maintaining utility.
Abstract:We consider persistent LLMs that accumulate memories of their interactions with a user over time. Such LLMs maintain memories using external storage, which they can query to overcome the limitations of a fixed context window. Such systems have numerous practical applications, as they can draw on all past interactions when responding to user queries. In this paper, we ask whether LLMs can forget information shared with them upon a user's request. We find that current LLMs fail to delete such information---even when they claim to have forgotten it and even when operating with a limited context. To this end, we consider a new direction of study: Deletion of LLM Memories. We show that naively removing messages that match a user's deletion request is insufficient, since conversations naturally introduce message dependencies that cause information to persist. To correctly handle deletion requests, we propose the DeLLM framework. It dynamically constructs relevant context for each LLM query and maintains a provenance graph of messages to determine which ones must be removed during deletion. Our experiments show that DeLLM achieves a high deletion rate while maintaining utility.
Computation and Language (cs.CL)
arXiv:2609.36414 [cs.CL]
(or arXiv:2609.36414v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2609.36414 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 278 cuvinte
Subiecte conexe
What's Continual Learning, and Why Might We Expect To See It In Advanced LLM Agents?
LessWrong · 12 iun. 2026
Links #3: 2026/06 Part 1
LessWrong · 15 iun. 2026
The Five Realities of Existence, by Isaac Megbolugbe
promptnewsonline.com · 6 sept. 2026
How might continual learning affect safety and alignment?
LessWrong · 13 iun. 2026