Prompt injection: analysis of 11 November 2026
Back to the Chronology · Current news in this section

Analysis
Prompt injection has moved from a research curiosity to real attacks on AI agents, and the security industry is spending heavily to defend against it.
Over the past month, reporting shows prompt injection and jailbreaks causing real-world harm. A jailbroken AI agent reportedly ran a two-week attack on a European telecom, researchers found adware hidden in an OpenAI Codex jailbreak toolkit, and a Brazilian court fined lawyers for hiding a prompt injection in a filing. The defence market is consolidating: CrowdStrike is buying Fiddler and Dreadnode for about $495 million, and Palo Alto Networks is in talks to buy Clutch for $500 million. Researchers keep finding injections in new places, from HTTP headers to PDFs and API gateways. A steady stream of papers and practitioner posts argues that classifiers alone cannot stop the attacks and that agents need system-level controls.
Key developments
- Attacks in the wild. A jailbroken agent ran a two-week attack on a European telecom, and a Codex jailbreak toolkit hid prompt adware. 311
- Courts meet prompt injection. A Brazilian court fined lawyers $16,000 for hiding a prompt injection in a court filing. 1415
- Big money for agent security. CrowdStrike is paying about $495 million for Fiddler and Dreadnode, Palo Alto is in talks to buy Clutch for $500 million, and OpenAI engaged Hacktron to test its systems. 1213109
- New injection surfaces. A study of 217 million web responses found injections in HTTP headers, others hide them in PDFs and API gateways, and a dataset tracks prompt-injection CVEs. 257176
- Research: agents as untrusted components. Papers scale attacks with agent swarms, measure how harnesses affect web agents, and propose deterministic policies and least-privilege auditing of tool calls. 31303235
- Benchmarks and standards. DeepMind released Gemini Robotics 2.5 with an open ASIMOV-jailbreak benchmark, and OWASP lists prompt injection as the top LLM risk. 123
Themes
Stories this analysis is based on (22)
- DeepMind reveals Gemini Robotics 2.5 and open safety benchmark "ASIMOV-jailbreak" Ars Technica
- Jailbroken AI Agent Ran Two-Week Attack on European Telecom Infosecurity Magazine
- Open dataset of prompt-injection CVEs with a monthly threat index Hacker News
- Hiding prompt injections in PDFs, and six judges that catch them Hacker News
- OpenAI taps AI security startup Hacktron to test systems Tech in Asia
- Palo Alto Networks in talks to buy AI agent security startup Clutch for $500M TechCrunch
- Researchers Uncover Prompt Adware Hidden in OpenAI Codex Jailbreak Toolkit The Hacker News
- CrowdStrike to shell out nearly $500M for 2 AI security startups as the market for securing agents takes off TechCrunch
- CrowdStrike Reportedly Paying $495 Million for AI Security Firms Fiddler, Dreadnode SecurityWeek
- Brazilian lawyers fined $16K for hiding a prompt injection in a court filing Hacker News
- Brazilian Court Sanctions Lawyers for Prompt Injection in Court Filing Hacker News
- Who Governs the Governor? – Stored Prompt Injection Through the API Gateway Hacker News
- Prompt injection scanners should be positional, not phrasal Hacker News
- LLM01: 2025 Prompt Injection Hacker News
- Prompt injection in HTTP headers: what 217M web responses show Hacker News
- Prompt injection classifiers are terrible security boundaries Hacker News
- Harness Matters: Measuring and Dissecting Security Resilience in Web Agents arXiv (research papers)
- Scaling Adaptive Prompt Injection Attacks with Agent Swarms arXiv (research papers)
- Treating the Model as an Untrusted Component: Deterministic Policy Enforcement, Human Approval and Taint Tracking Against Prompt Injection in Tool-Using Agents, with Validation on Open-Weights Models arXiv (research papers)
- When Coding Agents Get Arms: Safety Risks Across the Digital-Physical Boundary arXiv (research papers)
- Least-Privilege Auditing of Agent Tool Calls with Intent Graphs and Reasoning-Guided Escalation arXiv (research papers)
- Agents of Chaos: A Systematization of Knowledge (SoK) on Malicious Attacks on Agentic AI Systems arXiv (research papers)
Written by Claude (Anthropic), based on 41 recent headlines and excerpts in this section, Nov 11. AI analysis can misread or miss things: the numbers link to the original reporting, which is the authority.