Practitioner-grade analysis of offensive AI security. Prompt injection, model jailbreaks, agent and tool-use exploitation, AI red team techniques, and adversarial ML — distilled from primary sources, not press releases.https://aisec.blog/enhttps://aisec.blog/posts/llm-attack/https://aisec.blog/posts/llm-attack/A practitioner's map of LLM attack classes — from direct prompt injection and jailbreaks to indirect injection, RAG poisoning, and agent tool-call abuse — organized by OWASP 2025 and MITRE ATLAS.Mon, 22 Jun 2026 00:00:00 GMTprompt-injectionjailbreakagent-securityadversarial-mlred-teamllm-securityAI Sec Editorialhttps://aisec.blog/posts/prompt-injection-examples-2/https://aisec.blog/posts/prompt-injection-examples-2/Concrete prompt injection examples across five attack classes — direct override, system-prompt leak, indirect RAG poisoning, agent tool-call hijack, and multimodal smuggling — with PoC payloads and defender actions.Sun, 21 Jun 2026 00:00:00 GMTprompt-injectionattack-techniquesllm-securityred-teamagent-securityindirect-injectionAI Sec Editorialhttps://aisec.blog/posts/llm-bypass-2/https://aisec.blog/posts/llm-bypass-2/A practitioner map of LLM bypass technique families — prompt injection, jailbreak personas, encoding obfuscation, RAG poisoning, and agent-specificSat, 13 Jun 2026 00:00:00 GMTllm-bypassprompt-injectionjailbreakadversarial-mlred-teamagent-securityAI Sec Editorialhttps://aisec.blog/posts/ai-red-team/https://aisec.blog/posts/ai-red-team/A practitioner's guide to AI red teaming — what makes LLM attack surface different from traditional app testing, the techniques that reliably produceFri, 05 Jun 2026 00:00:00 GMTred-teamprompt-injectionjailbreakagent-securityllm-securitytoolingAI Sec Editorialhttps://aisec.blog/posts/prompt-hacking-2/https://aisec.blog/posts/prompt-hacking-2/Prompt hacking covers three distinct attack classes against LLMs: direct injection, indirect injection, and jailbreaking.Tue, 02 Jun 2026 00:00:00 GMTprompt-injectionjailbreakred-teamllm-securityagent-securityAI Sec Editorialhttps://aisec.blog/posts/adversarial-ml-attack-taxonomy-red-team-reference/https://aisec.blog/posts/adversarial-ml-attack-taxonomy-red-team-reference/A working taxonomy of attacks against ML systems — evasion, poisoning, privacy, and abuse — mapped to attacker knowledge and capability, grounded in theSat, 23 May 2026 00:00:00 GMTadversarial-mlred-teamevasionpoisoningmodel-extractionmembership-inferencetaxonomyAI Sec Editorialhttps://aisec.blog/posts/ai-red-team-engagement-methodology-scoping-to-reporting/https://aisec.blog/posts/ai-red-team-engagement-methodology-scoping-to-reporting/The full lifecycle of an LLM red team engagement — scoping and rules of engagement, threat modeling, the test plan by attack class, the tooling that runsSat, 23 May 2026 00:00:00 GMTred-teammethodologyllm-securitytoolingscopingreportinggarakpyritAI Sec Editorialhttps://aisec.blog/posts/position-behavioural-assurance-cannot-verify-the-safety-clai/https://aisec.blog/posts/position-behavioural-assurance-cannot-verify-the-safety-clai/A new position paper by Seth and Sankarapu formalizes the structural mismatch between what AI governance frameworks require evaluators to verify and whatSat, 16 May 2026 00:00:00 GMTred-teamadversarial-mlagent-securitygovernancebehavioral-evaluationinterpretabilityAI Sec Editorialhttps://aisec.blog/posts/understanding-prompt-injections-a-frontier-security-challeng/https://aisec.blog/posts/understanding-prompt-injections-a-frontier-security-challeng/OpenAI's November 2025 advisory on prompt injection arrived the same week a 14-researcher arXiv paper showed adaptive attacks achieve >90% success againstSat, 16 May 2026 00:00:00 GMTprompt-injectionindirect-injectionagent-securityllm-securityred-teamAI Sec Editorialhttps://aisec.blog/posts/llm-prompt-injection-2/https://aisec.blog/posts/llm-prompt-injection-2/A practitioner's breakdown of how LLM prompt injection payloads are constructed, why the threat class changes when agents can invoke tools, and whatThu, 14 May 2026 00:00:00 GMTprompt-injectionagent-securityred-teampayload-constructionindirect-injectionAI Sec Editorialhttps://aisec.blog/posts/prompt-injection-attack-2/https://aisec.blog/posts/prompt-injection-attack-2/Unit 42 documented 12 prompt injection attacks in production with 22 distinct delivery techniques. Here's how attackers build payloads that reach theThu, 14 May 2026 00:00:00 GMTprompt-injectionred-teamattack-techniquesllm-securitypayload-deliveryAI Sec Editorialhttps://aisec.blog/posts/faq-llm-security-fundamentals/https://aisec.blog/posts/faq-llm-security-fundamentals/Three essential questions for anyone building, securing, or red-teaming LLM applications — covering the distinction between jailbreaks and promptTue, 12 May 2026 00:00:00 GMTfaqprompt-injectionjailbreakllm-securityred-teamdefenseAI Sec Editorialhttps://aisec.blog/posts/prompt-injection-examples/https://aisec.blog/posts/prompt-injection-examples/A technical breakdown of real prompt injection examples — direct, indirect, multimodal, and RAG-poisoning attacks — with conditions, payloads, and whatTue, 12 May 2026 00:00:00 GMTprompt-injectionllm-securityred-teamagent-securityadversarial-mlAI Sec Editorialhttps://aisec.blog/posts/agent-tool-use-exfiltration/https://aisec.blog/posts/agent-tool-use-exfiltration/Why agentic LLM systems convert injection bugs into data exfiltration, financial loss, and remote code execution — with concrete attack chains and theMon, 11 May 2026 00:00:00 GMTagentsprompt-injectiontool-usespokellm-securityAI Sec Editorialhttps://aisec.blog/posts/ai-red-team-hub/https://aisec.blog/posts/ai-red-team-hub/The central resource index for offensive AI security on aisec.blog — prompt injection, jailbreaks, adversarial ML, red team methodology, and toolingMon, 11 May 2026 00:00:00 GMThubred-teamprompt-injectionjailbreakadversarial-mlllm-securitytoolingAI Sec Editorialhttps://aisec.blog/posts/direct-vs-indirect-prompt-injection/https://aisec.blog/posts/direct-vs-indirect-prompt-injection/Direct and indirect prompt injection are fundamentally different attacks with different attack surfaces, threat actors, and mitigations.Mon, 11 May 2026 00:00:00 GMTprompt-injectionindirect-injectionattack-vectorsllm-securitythreat-modelingapplication-securityAI Sec Editorialhttps://aisec.blog/posts/indirect-injection-rag-pipelines/https://aisec.blog/posts/indirect-injection-rag-pipelines/How retrieval-augmented generation surfaces become injection vectors, with concrete attack patterns from production RAG systems and the chunkingMon, 11 May 2026 00:00:00 GMTprompt-injectionragindirect-injectionspokellm-securityAI Sec Editorialhttps://aisec.blog/posts/jailbreak-ai/https://aisec.blog/posts/jailbreak-ai/A technical guide to jailbreak AI attacks — from manual prompt exploits to automated adversarial suffixes — covering the major technique familiesMon, 11 May 2026 00:00:00 GMTjailbreakadversarial-mlred-teamllm-securityprompt-injectionAI Sec Editorialhttps://aisec.blog/posts/jailbreak-llm/https://aisec.blog/posts/jailbreak-llm/How automated jailbreak LLM techniques like TAP use attacker LLMs to iteratively crack target models, why success transfers across model families, andMon, 11 May 2026 00:00:00 GMTjailbreakllm-securityred-teamadversarial-mlautomated-attacksAI Sec Editorialhttps://aisec.blog/posts/llm-bypass/https://aisec.blog/posts/llm-bypass/A technical breakdown of LLM bypass techniques — adversarial suffixes, shallow alignment exploits, fine-tuning attacks, and guardrail evasion — withMon, 11 May 2026 00:00:00 GMTjailbreakllm-securityred-teamadversarial-mlalignmentAI Sec Editorialhttps://aisec.blog/posts/llm-jailbreak/https://aisec.blog/posts/llm-jailbreak/A technical breakdown of LLM jailbreak attack classes — many-shot, Crescendo multi-turn escalation, roleplay, and encoding tricks — plus an honest look atMon, 11 May 2026 00:00:00 GMTjailbreakllm-securityred-teamprompt-engineeringadversarial-mlAI Sec Editorialhttps://aisec.blog/posts/llm-prompt-injection/https://aisec.blog/posts/llm-prompt-injection/A technical breakdown of LLM prompt injection — direct, indirect, and agent-targeting variants — grounded in real-world attack patterns observed inMon, 11 May 2026 00:00:00 GMTprompt-injectionllm-securityagent-securityred-teamindirect-injectionAI Sec Editorialhttps://aisec.blog/posts/model-extraction-vs-model-inversion/https://aisec.blog/posts/model-extraction-vs-model-inversion/Model extraction and model inversion both threaten model confidentiality, but they target different aspects of the model and require different defenseMon, 11 May 2026 00:00:00 GMTmodel-extractionmodel-inversionmodel-theftmembership-inferencetraining-data-privacyllm-securityattack-vectorsAI Sec Editorialhttps://aisec.blog/posts/prompt-hacking/https://aisec.blog/posts/prompt-hacking/A practitioner's breakdown of prompt hacking — the three attack families (injection, leaking, jailbreaking), how each works mechanically, and whatMon, 11 May 2026 00:00:00 GMTprompt-injectionjailbreakred-teamllm-securityadversarial-mlAI Sec Editorialhttps://aisec.blog/posts/prompt-injection-attack-compendium/https://aisec.blog/posts/prompt-injection-attack-compendium/A practitioner's pillar reference on prompt injection attacks against LLM systems — direct and indirect variants, real-world payloads, detection signalsMon, 11 May 2026 00:00:00 GMTprompt-injectionllm-securityai-red-teampillarowasp-llm01AI Sec Editorialhttps://aisec.blog/posts/prompt-injection-attack/https://aisec.blog/posts/prompt-injection-attack/A practitioner's breakdown of prompt injection attacks — direct, indirect, and multi-modal — covering the HouYi framework, real CVEs, and mitigations thatMon, 11 May 2026 00:00:00 GMTprompt-injectionred-teamllm-securityagent-securityadversarial-mlAI Sec Editorialhttps://aisec.blog/posts/prompt-injection-detection-signals/https://aisec.blog/posts/prompt-injection-detection-signals/The observable signals that indicate a prompt injection attempt or success in a live LLM application — input classifiers, output classifiers, canaryMon, 11 May 2026 00:00:00 GMTprompt-injectiondetectionllm-monitoringspokellm-securityAI Sec Editorialhttps://aisec.blog/posts/gpt4-jailbreak/https://aisec.blog/posts/gpt4-jailbreak/Three active attack classes — IRIS self-refinement, Crescendo multi-turn escalation, and classic prompt-engineering patterns — consistently breach GPT-4Sun, 10 May 2026 00:00:00 GMTjailbreakgpt-4prompt-engineeringred-teamllm-securityAI Sec Editorialhttps://aisec.blog/posts/llm-security/https://aisec.blog/posts/llm-security/What LLM security actually means in 2026 — the attack classes red teamers test, the controls that hold up under fire, and the frameworks that map the territory.Sat, 09 May 2026 00:00:00 GMTllm-securityprompt-injectionred-teamowaspagent-securityAI Sec Editorialhttps://aisec.blog/posts/prompt-injection-bypass-classes/https://aisec.blog/posts/prompt-injection-bypass-classes/Vendor 'AI guardrails' detect 80% of textbook payloads and 30% of real ones. Here's how attackers actually bypass them — and what your detection layer isThu, 07 May 2026 00:00:00 GMTprompt-injectionred-teamguardrailsllm-securitybypass-techniquesAI Sec Editorialhttps://aisec.blog/posts/ai-jailbreak/https://aisec.blog/posts/ai-jailbreak/An AI jailbreak is any input that makes an aligned language model violate its own safety policy. We walk through the technique families that actuallyWed, 06 May 2026 00:00:00 GMTjailbreakprompt-injectionllm-securityred-teamadversarial-mlAI Sec Editorialhttps://aisec.blog/posts/chatgpt-jailbreak-prompt/https://aisec.blog/posts/chatgpt-jailbreak-prompt/A research-grounded breakdown of ChatGPT jailbreak prompt categories — DAN, privilege escalation, persona injection, and multi-turn escalation — plus whatTue, 05 May 2026 00:00:00 GMTjailbreakprompt-injectionred-teamchatgptllm-securityadversarial-mlAI Sec Editorialhttps://aisec.blog/posts/anyone-wanna-learn-the-ceh-or-oscp-red-teaming-free/https://aisec.blog/posts/anyone-wanna-learn-the-ceh-or-oscp-red-teaming-free/A Reddit offer to teach OSCP and CEH fundamentals for free surfaces a question every traditional pentester should answer: which of those skills transferSun, 03 May 2026 00:00:00 GMTred-teamoscpcehtoolingprompt-injectionadversarial-mlAI Sec Editorialhttps://aisec.blog/posts/flashrt-towards-computationally-and-memory-efficient-red-tea/https://aisec.blog/posts/flashrt-towards-computationally-and-memory-efficient-red-tea/A new optimization-based red-teaming framework claims 2–7x speedup and 2–4x lower memory than nanoGCG against 32K-context LLMs, putting GCG-class attacksSun, 03 May 2026 00:00:00 GMTprompt-injectionred-teamgcglong-contextknowledge-corruptionragAI Sec Editorial