CheeseDB Introduces Revocable Memory to Solve AI Data Poisoning
A new open-source database engine for AI agents allows developers to purge specific memories by deleting their original data sources.
Photo: Hacker News Front Page
Key points
- CheeseDB enables provenance-aware memory, allowing AI to forget information when a source is revoked.
- The engine is written in native Go and integrates with the Model Context Protocol (MCP).
- Benchmarks indicate the system outperforms traditional SQL and graph databases for specific AI memory management.
A new database engine called CheeseDB has been released in preview, designed specifically to manage how AI agents store and recall information. Unlike traditional databases, this system tracks the "provenance" of every memory. This allows developers to delete a specific source—such as a compromised web page or an incorrect user message—and automatically trigger the removal of every fact the AI learned from that specific input.
Built using the Go programming language, CheeseDB is an embedded, hash-chained memory store. It addresses a common failure in current AI systems where agents continue to act on outdated or poisoned information even after the original document has been removed from their knowledge base. The engine is currently compatible with the Model Context Protocol (MCP), allowing it to be used with AI-powered development tools like Cursor and Claude Code.
Technical benchmarks conducted on Apple silicon suggest that the engine provides faster data retrieval and deletion for AI-specific tasks compared to standard SQL or graph databases like PostgreSQL and Neo4j. However, the developers emphasize that the project is still in active development and has not yet undergone an independent security review, making it unsuitable for production environments at this stage.
What to watch next: The potential integration of provenance-aware memory into mainstream agent frameworks to improve data reliability and safety.Source Hacker News Front Page — github.com/govinargo-labs/cheesedb ↗
Read next
🚀Startups & FundingHarvard Study Reveals US Hyperscale Data Centers Consume Power Rivaling Entire States
A new Harvard study reveals that America's 403 hyperscale data centers consumed between 68 and 99 terawatt-hours of electricity during the 2024-25 study period, rivaling …
🚀Startups & FundingCalifornia Files Lawsuit Against OpenAI Over Autonomous AI Incidents
California Attorney General Rob Bonta has filed a lawsuit against OpenAI, alleging the company failed to control its autonomous AI agents and misled the public about safe…
🏢Big TechUnited Launch Alliance Investigates Vulcan Rocket Failure
United Launch Alliance is investigating a mission failure of its Vulcan rocket, which suffered from low booster performance shortly after liftoff. The incident raises que…
🌐Work & SocietyUS Southwest Hydropower Output Hits Record Lows Amid Persistent Drought
Hydropower generation in the US Southwest dropped 28% below the 10-year average this summer as record-low water levels in the Colorado River Basin threaten regional energ…
🏢Big TechYouTube Updates View Metrics to Include TV Co-Viewing Estimates
YouTube is incorporating estimated co-viewing data into its public view counts to better reflect group consumption on television sets.
Apple Services Experience Widespread Outage
Apple users are reporting significant disruptions across iCloud, the App Store, and other core services, despite initial discrepancies in the company's status reporting.