Knowledge Base Regulation Tracker Standards & Frameworks OSS Directory AI Incident Database Research Paper Library Case Studies Society Articles News Jobs About Charter Sign inBecome a Member
Home/Research Paper Library/Persistent Pre-training Poisoning of LLMs

Persistent Pre-training Poisoning of LLMs

Showed that pretraining-time poisoning of a tiny data fraction persists through alignment, motivating the near-constant-samples follow-up.

pretraining poisoningpersistence

Research question

Do poisoning effects introduced during pretraining survive post-training alignment?

Key findings

  • Poisoning 0.1% of pretraining data implants behaviours that persist through SFT and DPO - Denial-of-service and belief-manipulation attacks persist; jailbreak backdoors are less durable

Sources

PreviousImprompterNextAgentHarm

Contribute

Spot an error, or know a record we should add? Members can propose additions and submit corrections.

Submit researchReport a correction or request removal

Related in Research Paper Library

The Price of Safety: Benign-Case Utility and Token Overhead of Memory-Poisoning Defenses in LLM Agents

Memory-poisoning defenses for LLM agents are typically evaluated by their ability to prevent attacks. However, the traffic they process is rarely adversarial. The cost of implementing a defense is paid with each…

Data poisoning2026

Micro-Collaborative Poisoning: A Distributed Attack on RAG Systems

Retrieval-Augmented Generation (RAG) improves large language models by grounding outputs in external knowledge sources, but this dependency also creates a surface for poisoning attacks. This paper introduces…

Data poisoning2026

PentestChain: A Cost-Aware, MCP-Orchestrated Framework for Automated Penetration Testing with Free-Tier LLMs

AI-driven penetration testing has been demonstrated with premium frontier models such as GPT-4, but the per-engagement token cost makes continuous, automated testing unaffordable for the smaller organisations that…

Data poisoning2026