Spotlighting

Prompting techniques that mark untrusted input so the model can tell data from instructions.

Records citing Spotlighting

Sep 29, 2026
ToolFence proposes typed capabilities with parameter provenance to authorize agent tool calls, tested on AgentDojo
DefensePaperHong Kong Polytechnic University, Shandong University

Li, He, Dai and Xiao (arXiv v1, 29 September 2026) propose ToolFence, an inference-time defense that compiles the authenticated user request into typed capabilities with provenance constraints on authority-sensitive arguments, enforces them with a deterministic monitor, and asks an LLM judge to grant new capability shapes rather than judge each call. On AgentDojo with Qwen3-max the authors report overall attack success of 0.20% against 21.20% undefended, with clean utility 38.90% against 42.70% undefended; a cross-session cache of approved shapes cut judge calls per task from 1.84 to 1.05 in their ablation. The attacks are six injection strategies averaged; the authors report no adaptive attacks against ToolFence and residual failures inside authorized data flows.

Jun 11, 2025
LLMail-Inject releases data from an adaptive prompt injection challenge against an email agent
DefenseDatasetMicrosoft

Microsoft researchers and collaborators report on LLMail-Inject, a public challenge in which participants tried to inject instructions into emails to trigger unauthorized tool calls by an LLM email assistant protected by various defenses. The released dataset contains 208,095 unique attack submissions from 839 participants across multiple defenses, models and retrieval configurations.

May 20, 2025
Google DeepMind reports lessons from continuously attacking Gemini with adaptive prompt injections
DefensePaperGoogle DeepMind

Shi and colleagues describe Google DeepMind's continuous adaptive-attack evaluation of Gemini against indirect prompt injection in tool-use settings. On Gemini 2.0, adaptive attacks generally matched or beat non-adaptive ones against eight baseline defenses, reaching 98.4% against in-context learning and 82.4% against spotlighting, while a warning defense and a user-instruction classifier held (at most 10.8% and 3.0%). Adversarial fine-tuning for Gemini 2.5 lowered but did not eliminate attack success.

Mar 20, 2024
Microsoft researchers propose spotlighting to mark untrusted input against indirect prompt injection
DefensePaperMicrosoft

Hines and colleagues at Microsoft describe spotlighting, a family of prompt-engineering transformations (delimiting, datamarking, encoding) that signal to the model where untrusted text came from. On GPT-family models they report attack success falling from above 50% to under 2% with minimal task impact. Microsoft later described spotlighting as one layer of its production defense-in-depth.