A developer built buried-injections, an open-source benchmark that buries 629 real indirect prompt-injection attacks from ETH Zürich's AgentDojo inside ordinary tool output and runs ten open-source de…
A security engineer published a practical checklist for testing tool-calling LLM agents, arguing that test suites must verify the system of record rather than the agent's final reply. The checklist co…
Prompt injection ranks #1 on the OWASP LLM Top 10 (OWASP LLM01), and Google measured a +32% jump in indirect prompt injection in the web corpus between November 2025 and February 2026, according to Th…
A developer built taintgate, a provenance-tracking gate that sits between an AI agent and its tools to block prompt-injection attacks that detectors miss. Testing 10 free open-source detectors against…
A new arXiv paper (2609.05992v1) introduces Multi-Objective Agent Evolution (MOAE), a Pareto-preserving evolutionary search method that optimizes LLM-based agents across task performance, trajectory q…
A researcher revising a paper on benchmark evaluation frameworks acknowledged that the current arXiv version lags behind the working manuscript, which now introduces a claim-relative inference layer s…
A new authorization framework called the Agentic Principal Chain (APC) reduces data exfiltration in multi-agent AI systems from 75-100% to 0% across all four AgentDojo domains and blocks all 544 Injec…
Crawdad, an AI agent security proxy, published Contemporary Agent Attacks, an open benchmark of 497 attack samples and 1,172 benign ones, including a documented failure list where its own product scor…
Uber Technologies Inc. has open-sourced its enterprise security system for AI agents, called Agentic AI Detection and Response (ADR), which tracks and blocks malicious agent actions. The software, use…
A developer has open-sourced a framework-agnostic testing methodology for AI agents, including a 61-source benchmark map and 58 universal test blocks across seven tiers, with full coverage of the OWAS…
A pre-deployment hardening pipeline that scans prompt templates and tool interfaces can eliminate basic data leakage entirely and reduce high-stress manipulation attacks by 91%, according to testing o…
Researchers propose Twin Agent, a privilege separation design pattern for LLM agents that uses two nearly symmetric agents—an Explore Agent inspecting untrusted information and a Safe Agent executing …
LLMbda, a new framework based on untyped call-by-value lambda calculus, introduces a provenance-based approach to defend AI agents against prompt injection attacks. In tests on the AgentDojo banking b…
An independent researcher found that existing indirect prompt injection benchmarks like BIPIA, InjecAgent, and AgentDojo may produce unreliable scores due to reliance on LLM-judges and evaluation of e…
Guardian, an open-source user-space firewall for AI agents, has released v0.1.0, intercepting and evaluating agent actions with a deterministic policy engine. In testing, it reduced prompt-injection a…
A new benchmark, dspy-security-bench, reveals that DSPy prompt optimization degrades adversarial robustness against harder prompt-injection attacks. Testing with AgentDojo's attack suite, optimizers l…