Deleted AI Chats Leave Memory Traces That Can Be Recovered, Georgia Tech Researchers Find

Local AI Assistants Leave Recoverable Memory Traces

Deleting a conversation with local AI assistants on your phone may not erase everything. Researchers at Georgia Tech have developed ORISA, a tool that can recover deleted conversations from AI assistants running on Android phones and rebuild the recovered information into a working AI session1

1

. The findings reveal that deleted AI chats don't necessarily remove every trace, particularly when local LLMs process conversations directly on devices rather than through cloud services.

Local AI assistants run directly on a phone rather than sending conversations to a cloud service. While the local assistant is running, information from conversations can remain in the phone's working memory even after deletion2

2

. ORISA searches this memory for pieces of deleted conversations, exploiting residual data that lingers in active memory.

Data Recovery Exceeds Original Session Capabilities

In one experiment, researchers created a fictional AI assistant called "Project Note Keeper" using the ChatterUI Android app. They fed it 100 fictional facts about a project, then deleted the conversation1

1

. ORISA recovered information from the deleted conversation and used it to reconstruct conversations into a new working session. When researchers asked the restored assistant to list the project notes it had been given, it returned 26 facts. Remarkably, the original session could access only 14 facts because the rest had fallen outside its active memory2

2

.

The experiment demonstrated that researchers could recover deleted conversations containing information the AI had already stopped actively using. "It can help establish what information someone shared with an AI assistant and what assistance they received," said Ph.D. student Ken Xu, lead author of the study2

2

.

Broad Testing Confirms Recovery Across Multiple Platforms

The researchers tested ORISA with 10 Android apps and three AI models. Across 60 test configurations, the tool recovered conversation information that remained in the phone's memory1

1

. The researchers note that the information becomes harder to recover as the phone's working memory is overwritten by other activity, suggesting a time-sensitive window for data recovery.

Privacy Concerns for Users and Opportunities for Digital Investigators

The research presents a dual-edged sword. It could give digital investigators a new way to recover information from phones, advancing digital forensics capabilities2

2

. However, it also raises significant privacy concerns for users who use AI assistants that run locally on their devices, believing their deleted conversations are truly gone. The discovery of these memory traces challenges assumptions about data permanence on local devices.

The research was presented at the 2026 IEEE Symposium on Security and Privacy. The paper, titled "Recovering and Rehosting Mobile Local LLM Conversations and Contexts via Memory Forensics," was authored by Ph.D. students Xu, David Oygenblik, Runze Zhang, and Mingxuan Yao; Research Scientist Muhammad Ibrahim; and Associate Professor Brendan Saltaformaggio2

2

. The work was done in Georgia Tech's Cyber Forensics Innovation Laboratory, a collaboration between the School of Cybersecurity and Privacy and the School of Electrical and Computer Engineering.

What This Means for Privacy Risks Going Forward

As local AI assistants become more prevalent on Android phones and other devices, users should understand that deletion doesn't guarantee complete erasure. The privacy risks extend beyond immediate concerns—law enforcement and digital investigators now have tools to access supposedly deleted information. Users relying on local processing for enhanced privacy may need to reconsider their threat models. Watch for potential updates from AI assistant developers addressing these memory traces, and expect regulatory discussions around data retention standards for local LLMs.