Securing AI Agents Against Prompt Injection Attacks
Nov 2025 benchmark of 847 adversarial test cases across 5 attack categories against 7 LLMs. The combined defense framework reduces attack success from 73.2% to 8.7% while retaining 94.3% of baseline task performance.
- from
- Prompt Injection
- added
- 2026-10-10
- likes
- 0
similar
-
The Landscape of Prompt Injection Threats in LLM Agents (SoK) arxiv.org
Feb 2026 systematization-of-knowledge paper with a unified taxonomy covering attack payload strategies (heuristic vs. optimisation-based) and defense intervention stages (text, model, execution). Introduces the AgentPI benchmark for context-dependent agent tasks that all prior…
-
Attention Tracker: Detecting Prompt Injection Attacks in LLMs aclanthology.org
NAACL 2025 Findings paper detecting prompt injection by tracking attention distribution shifts, with no modification to the underlying model required, making it deployable as a wrapper on any LLM.
-
New Prompt Injection Papers: Agents Rule of Two and The Attacker Moves Second simonwillison.net
Simon Willison's Nov 2025 commentary on both landmark papers, including the finding that 12 published defenses were bypassed at >90% success rate using gradient descent and RL-based adaptive attacks.
-
Design Patterns for Securing LLM Agents against Prompt Injections simonwillison.net
Overview of various strategies to mitigate the risk of prompt injection.
-
Prompt Injection 2.0: Hybrid AI Threats arxiv.org
Jul 2025 paper showing how prompt injections now combine with XSS, CSRF, AI worm propagation, and multi-agent infections to evade traditional WAFs entirely. Evaluates Preamble's classifier, data-tagging, and RL-based defenses against these hybrid scenarios.
-
How AI Prompt Injection Works | Hands-on with LLMs youtube.com
Jan 2026 AppSecEngineer tutorial with a code-level demo of injecting against a real LLM application and live testing of LLM Guard detection. One of the most practical end-to-end tutorials published to date.
Prompt Injection › Research Papers: “Nov 2025 benchmark of 847 adversarial test cases across 5 attack categories against 7 LLMs. The combined defense framework reduces attack success from 73.2% to 8.7% while retaining 94.3% of baseline task performance.”