AI Security Topics
Every offensive AI security topic covered here: prompt injection, jailbreaks, agent and tool-use exploitation, adversarial ML, and red team methodology.
Tags
- #red-team 19
- #llm-security 18
- #prompt-injection 18
- #agent-security 11
- #adversarial-ml 10
- #jailbreak 8
- #tooling 3
- #attack-vectors 2
- #indirect-injection 2
- #long-context 2
- #owasp 2
- #tool-use 2
- #adversarial-training 1
- #agents 1
- #ai-red-team 1
- #alignment 1
- #application-security 1
- #attack-techniques 1
- #automated-attacks 1
- #behavioral-evaluation 1
- #bypass-techniques 1
- #ceh 1
- #chatgpt 1
- #custom-gpts 1
- #cve 1
- #defense 1
- #defenses 1
- #evasion 1
- #faq 1
- #garak 1
- #gcg 1
- #governance 1
- #gpt-security 1
- #guardrails 1
- #hub 1
- #indirect-prompt-injection 1
- #insecure-output-handling 1
- #interpretability 1
- #knowledge-corruption 1
- #llm-bypass 1
- #llm-security-vulnerabilities 1
- #many-shot-jailbreaking 1
- #mcp 1
- #medical-ai 1
- #membership-inference 1
- #methodology 1
- #model-extraction 1
- #multi-turn 1
- #multimodal 1
- #openai 1
- #oscp 1
- #owasp-llm01 1
- #pillar 1
- #poisoning 1
- #prompt-engineering 1
- #pyrit 1
- #rag 1
- #reporting 1
- #scoping 1
- #spoke 1
- #taxonomy 1
- #threat-modeling 1
- #transferability 1
Categories
red-team 11 posts
- GPT Security: Attack Surfaces and Production ControlsA technical guide to GPT security covering prompt injection, custom GPTs, agent actions, data handling, and layered production controls.
- OSCP and CEH in 2026: What Carries Over to AI Red TeamingA free OSCP and CEH study offer raises the question every pentester should answer: which of those skills transfer to AI red teaming, and which do not.
- Insecure Output Handling Examples: XSS, SQLi, RCESee five insecure LLM output handling examples, from rendered XSS and query injection to code execution, plus the sink-side controls for each.
- LLM Security Vulnerabilities: What Actually Gets ExploitedA list of LLM security vulnerabilities: the OWASP 2025 Top 10, the CVEs where they turned real, and controls for injection and insecure output handling.
- GPT-Red: What OpenAI's Prompt-Injection Attacker ProvesOpenAI says GPT-Red beat human red teamers 84% to 13% on prompt injection. Here is what those numbers measure, and what they leave unanswered.
- LLM Attack Taxonomy: Injection, Jailbreaks, Agent AbuseA practitioner's map of LLM attack classes: direct and indirect prompt injection, jailbreaks, RAG poisoning, and agent tool-call abuse, mapped to OWASP.
prompt-injection 5 posts
- ChatGPT Exploits: Injection, Memory Hijack, Plugin AbuseHow ChatGPT gets exploited: indirect prompt injection, CSRF memory poisoning, cross-plugin request forgery, agent tool abuse, and the controls that help.
- FlashRT Cuts the GPU Bill on Long-Context Injection AttacksA new red-teaming framework claims 2-7x speedup and 2-4x lower memory than nanoGCG on 32K contexts, putting GCG-class attacks within a normal budget.
- Prompt Injection in 2025: OpenAI vs. Broken DefensesOpenAI's November 2025 injection advisory versus an October 2025 paper whose adaptive attacks broke most of 12 published defenses at over 90%.
- Prompt Injection Examples: A Practitioner's Attack LibraryA technical breakdown of real prompt injection examples across direct, indirect, multimodal, and RAG-poisoning attacks, with payloads and conditions.
- Prompt Injection Attack Compendium (2026 Edition)A pillar reference on prompt injection against LLM systems: direct and indirect variants, real payloads, detection signals, and the defenses that work.
jailbreak 4 posts
- Many-Shot Jailbreaking: How Long Context Breaks AlignmentMany-shot jailbreaking floods a long context with faux dialogues until refusal collapses. The power law behind it, why it transfers, and what blunts it.
- Automated Jailbreak Attacks and the Transfer ProblemHow automated attack generation works — PAIR, GCG, and TAP — why jailbreaks port across model families, and what that does to an assessment's threat model.
- LLM Bypass Techniques: Attacks on Alignment and GuardrailsLLM bypass techniques span adversarial suffixes, alignment-removing fine-tuning, and guardrail evasion, with defenses for each attack layer.
- LLM Jailbreak Techniques: Attack Taxonomy and DefensesLLM jailbreak attacks include roleplay, encoding, many-shot prompts, and multi-turn escalation, while layered controls limit their impact.
agent-security 3 posts
- APEX: Stopping Prompt Injection at the Execution BoundaryAPEX stops indirect prompt injection at the execution boundary with a pre-compiled task contract and decoy probes. 0% ASR on five of six benchmarks.
- ARSM: A Defense Pipeline for Medical LLM AgentsARSM-Agent chains six defense stages, from input risk perception to output control, for medical LLM agents. What each stage catches and where gaps remain.
- Agent Tool-Use Exfiltration: Attack Chains and ControlsIndirect prompt injection can make agents exfiltrate data, misuse tools, or execute code. Controls include least privilege, confirmation, and sandboxing.
primer 2 posts
- LLM Security FAQ: Prompt Injection, Jailbreaks, and DefensesThree foundational questions for anyone building or red-teaming LLM applications, covering jailbreaks, direct and indirect injection, and what defends.
- Direct vs. Indirect Prompt Injection: Threats and DefensesDirect and indirect prompt injection are fundamentally different attacks with different attack surfaces, threat actors, mitigations, and blast radius.