<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>LufSec Blog</title><description>Your Hub for Cybersecurity Training &amp; Research</description><link>https://blog.lufsec.com/</link><item><title>Attack Surface Intelligence: OWASP Amass Founder Jeff Foley on How Attackers Map Your Network</title><link>https://blog.lufsec.com/attack-surface-intelligence-owasp-amass-founder-jeff-foley-on-how-attackers-map-your-network/</link><guid isPermaLink="true">https://blog.lufsec.com/attack-surface-intelligence-owasp-amass-founder-jeff-foley-on-how-attackers-map-your-network/</guid><description>OWASP Amass founder Jeff Foley on attack surface intelligence, ownership attribution, AI in recon, and the career move most people miss.</description><pubDate>Wed, 11 Nov 2026 20:30:18 GMT</pubDate><category>Inside Cyber Minds</category><category>OWASP Amass</category><category>Attack Surface Management</category><category>OSINT</category><category>Recon</category><category>Jeff Foley</category><category>podcast</category><category>Cybersecurity</category></item><item><title>Anthropic&apos;s Glasswing Reveals the Guardrail Fallacy in AI Security</title><link>https://blog.lufsec.com/ai-security-threats-anthropic-glasswing-guardrail-fallacy/</link><guid isPermaLink="true">https://blog.lufsec.com/ai-security-threats-anthropic-glasswing-guardrail-fallacy/</guid><description>Anthropic&apos;s Project Glasswing found 129,000 flaws, proving that AI safety filters are not security. Learn why adversarial testing beats automated guardrails.</description><pubDate>Fri, 09 Oct 2026 13:08:00 GMT</pubDate><category>ai security</category><category>prompt injection</category><category>anthropic</category><category>adversarial testing</category></item><item><title>Automotive Hardware Attacks: An Architect&apos;s Guide to TARA</title><link>https://blog.lufsec.com/car-hacking-automotive-hardware-tara-guide/</link><guid isPermaLink="true">https://blog.lufsec.com/car-hacking-automotive-hardware-tara-guide/</guid><description>ISO/SAE 21434 is the automotive security standard, but its approach to hardware attack feasibility is fundamentally flawed. Here is why &apos;physical access&apos; is a dangerous blind spot for engineers.</description><pubDate>Thu, 08 Oct 2026 12:13:46 GMT</pubDate><category>car hacking</category><category>ISO 21434</category><category>TARA</category><category>automotive security</category><category>hardware hacking</category></item><item><title>The Cling Botnet and the Systemic Risk of IoT SDKs: Why You Need IoT Firmware Analysis</title><link>https://blog.lufsec.com/realtek-jungle-sdk-cling-botnet-iot-firmware-analysis/</link><guid isPermaLink="true">https://blog.lufsec.com/realtek-jungle-sdk-cling-botnet-iot-firmware-analysis/</guid><description>The Cling botnet&apos;s use of STUN for C2 is clever, but the real story is the Realtek Jungle SDK vulnerability. Learn why network scanning fails and how deep firmware analysis is the only way to stop systemic IoT threats.</description><pubDate>Wed, 07 Oct 2026 11:55:36 GMT</pubDate><category>iot security</category><category>firmware analysis</category><category>botnets</category><category>reverse engineering</category><category>realtek</category></item><item><title>BazaarBench: Why Your LLM Agent is a Financial Liability</title><link>https://blog.lufsec.com/ai-security-threats-bazaarbench-llm-delegation/</link><guid isPermaLink="true">https://blog.lufsec.com/ai-security-threats-bazaarbench-llm-delegation/</guid><description>LLM agents aren&apos;t just chatting; they&apos;re negotiating your bank balance and reputation. Discover why &apos;delegation safety&apos; is the new critical frontier in ai security threats.</description><pubDate>Tue, 06 Oct 2026 12:08:46 GMT</pubDate><category>ai security</category><category>prompt injection</category><category>llm agents</category><category>cybersecurity career</category></item><item><title>Containing the Autonomous Operator: Securing AI Agents on Kubernetes</title><link>https://blog.lufsec.com/securing-ai-agents-kubernetes-prompt-injection/</link><guid isPermaLink="true">https://blog.lufsec.com/securing-ai-agents-kubernetes-prompt-injection/</guid><description>When AI agents move from chat to infrastructure operations, prompt injection becomes an RCE vector. Learn how to secure the data-control boundary in Kubernetes.</description><pubDate>Mon, 05 Oct 2026 14:30:47 GMT</pubDate><category>ai security</category><category>kubernetes</category><category>prompt injection</category><category>devsecops</category><category>llmops</category></item><item><title>The Guardrail Paradox: How Specification Inconsistencies Power Prompt Injection</title><link>https://blog.lufsec.com/ai-security-threats-specification-inconsistencies-prompt-injection/</link><guid isPermaLink="true">https://blog.lufsec.com/ai-security-threats-specification-inconsistencies-prompt-injection/</guid><description>Your AI&apos;s safety guardrails aren&apos;t just porous—they are fighting a war against themselves. Learn how &apos;specification inconsistencies&apos; create the logical blind spots that make prompt injection possible.</description><pubDate>Fri, 02 Oct 2026 22:43:52 GMT</pubDate><category>ai security</category><category>prompt injection</category><category>LLM</category><category>Cybersecurity</category><category>Pentesting</category></item><item><title>ALIBI Attack: The AI Security Threat That Makes Malware Look Innocent</title><link>https://blog.lufsec.com/alibi-ai-security-threats-llm-malware-analyzers/</link><guid isPermaLink="true">https://blog.lufsec.com/alibi-ai-security-threats-llm-malware-analyzers/</guid><description>Researchers just showed how a fake README hidden in a binary can convince an LLM malware analyzer to clear real malware. ALIBI is prompt injection with a new suit on — and it changes what you need to test this week.</description><pubDate>Fri, 18 Sep 2026 12:46:46 GMT</pubDate><category>ai security</category><category>prompt injection</category><category>Malware Analysis</category><category>LLM Security</category><category>Threat Research</category></item><item><title>Car Hacking Starts at the Jenkins Server: What the South Korea APT Op Teaches Automotive Engineers</title><link>https://blog.lufsec.com/car-hacking-jenkins-server-south-korea-apt-automotive/</link><guid isPermaLink="true">https://blog.lufsec.com/car-hacking-jenkins-server-south-korea-apt-automotive/</guid><description>A North Korean APT just compromised South Korean automotive firms through load balancers — no CAN bus attack required. Here&apos;s the offensive skill gap it exposes for every automotive engineer.</description><pubDate>Thu, 17 Sep 2026 13:02:20 GMT</pubDate><category>car hacking</category><category>automotive security</category><category>APT</category><category>offensive security</category><category>supply chain</category></item><item><title>IoT Firmware Analysis: Why Signature Scans Miss Real Vulnerabilities</title><link>https://blog.lufsec.com/iot-firmware-analysis-semantic-vulnerability-detection/</link><guid isPermaLink="true">https://blog.lufsec.com/iot-firmware-analysis-semantic-vulnerability-detection/</guid><description>You unpacked the firmware. Strings ran. Nothing screamed. Here&apos;s why signature-based IoT firmware analysis fails on real embedded targets — and what to do instead.</description><pubDate>Wed, 16 Sep 2026 13:20:38 GMT</pubDate><category>iot security</category><category>firmware analysis</category><category>Vulnerability Research</category><category>reverse engineering</category><category>Embedded Security</category></item><item><title>AI Security Threats: What the Rogue-Agent Week Actually Means for Testers</title><link>https://blog.lufsec.com/ai-security-threats-rogue-agents-prompt-injection-testing/</link><guid isPermaLink="true">https://blog.lufsec.com/ai-security-threats-rogue-agents-prompt-injection-testing/</guid><description>Rogue AI agents and attacker-driven LLM exploit chains are exposing a testing gap most security teams didn&apos;t know they had. Here&apos;s how to close it without an ML degree.</description><pubDate>Tue, 15 Sep 2026 12:41:08 GMT</pubDate><category>ai security</category><category>prompt injection</category><category>LLM Security</category><category>Red Teaming</category><category>Weekly Recap</category></item><item><title>AI Security Threats: When Trusted AI Platforms Become the Attack Surface</title><link>https://blog.lufsec.com/ai-security-threats-trusted-ai-platforms-attack-surface/</link><guid isPermaLink="true">https://blog.lufsec.com/ai-security-threats-trusted-ai-platforms-attack-surface/</guid><description>Attackers are weaponizing Claude Artifacts, shared ChatGPT links, and sponsored AI search results as free malware CDNs. Here&apos;s what that means for your threat model.</description><pubDate>Mon, 14 Sep 2026 15:21:40 GMT</pubDate><category>ai security</category><category>prompt injection</category><category>LLM Security</category><category>Threat Intelligence</category><category>Application Security</category></item><item><title>Self-Issued Authentication: The AI Security Threat That Bypasses Your Prompt Injection Defenses</title><link>https://blog.lufsec.com/self-issued-authentication-ai-security-threats-llm/</link><guid isPermaLink="true">https://blog.lufsec.com/self-issued-authentication-ai-security-threats-llm/</guid><description>New research shows five major LLMs will design their own identity verification tests — and then accept trivial answers. A new class of AI security threats is here.</description><pubDate>Fri, 04 Sep 2026 12:27:33 GMT</pubDate><category>ai security</category><category>LLM Security</category><category>prompt injection</category><category>offensive security</category><category>Career</category></item><item><title>SPADE and the SPaT Attack: Why Signed V2X Messages Still Lie to Your Car</title><link>https://blog.lufsec.com/spade-spat-attack-car-hacking-v2x/</link><guid isPermaLink="true">https://blog.lufsec.com/spade-spat-attack-car-hacking-v2x/</guid><description>A September 2026 paper shows how attackers poison a connected vehicle&apos;s traffic-light feed after cryptographic authentication succeeds — and what car hacking skills fix the gap.</description><pubDate>Thu, 03 Sep 2026 11:14:26 GMT</pubDate><category>car hacking</category><category>V2X</category><category>connected vehicles</category><category>automotive security</category><category>Embedded Security</category></item><item><title>IoT Firmware Analysis: What the Switchvox Exploit Reveals About Embedded Attack Surface</title><link>https://blog.lufsec.com/iot-firmware-analysis-switchvox-embedded-attack-surface/</link><guid isPermaLink="true">https://blog.lufsec.com/iot-firmware-analysis-switchvox-embedded-attack-surface/</guid><description>The Switchvox CVE-2026-9586 exploit gives attackers a reverse shell without credentials. That&apos;s the entry point — IoT firmware analysis is what shows you the full blast radius.</description><pubDate>Wed, 02 Sep 2026 13:26:34 GMT</pubDate><category>iot security</category><category>firmware analysis</category><category>Vulnerability Research</category><category>Embedded Systems</category><category>Pentesting</category></item><item><title>AI Security Threats: What Anthropic&apos;s Forced Logout Teaches Career Builders</title><link>https://blog.lufsec.com/ai-security-threats-anthropic-infostealer-career/</link><guid isPermaLink="true">https://blog.lufsec.com/ai-security-threats-anthropic-infostealer-career/</guid><description>Anthropic just wiped payment data and force-logged thousands of Claude users after infostealer malware started harvesting AI sessions. Here&apos;s why it matters for your security career.</description><pubDate>Tue, 01 Sep 2026 12:55:15 GMT</pubDate><category>ai security</category><category>Malware Analysis</category><category>Infostealers</category><category>Career</category><category>Threat Intelligence</category></item><item><title>AI Security Threats Go Live: Infostealers Are Now Hijacking Claude Sessions</title><link>https://blog.lufsec.com/ai-security-threats-infostealer-claude-session-hijacking/</link><guid isPermaLink="true">https://blog.lufsec.com/ai-security-threats-infostealer-claude-session-hijacking/</guid><description>Anthropic just warned Claude users that infostealer malware is siphoning active sessions to drain usage. Here is what that attack actually looks like — and why it matters for anyone building a career in AI security.</description><pubDate>Mon, 31 Aug 2026 15:09:13 GMT</pubDate><category>ai security</category><category>Threat Intelligence</category><category>Infostealers</category><category>Session Hijacking</category><category>Cybersecurity Careers</category></item><item><title>Car Hacking Hits the Update Channel: What the Android Head-Unit Botnet Teaches Automotive Engineers</title><link>https://blog.lufsec.com/car-hacking-android-head-unit-update-hijack/</link><guid isPermaLink="true">https://blog.lufsec.com/car-hacking-android-head-unit-update-hijack/</guid><description>Click-fraud malware is landing on Android car head units through the same OTA channel manufacturers trust. Here&apos;s what automotive engineers are missing.</description><pubDate>Thu, 27 Aug 2026 11:27:30 GMT</pubDate><category>car hacking</category><category>automotive security</category><category>Android</category><category>Malware</category><category>OTA Updates</category><category>Embedded Security</category></item><item><title>AI Security Threats Just Got Personal: Inside the NVIDIA NeMo Webpage Attack</title><link>https://blog.lufsec.com/ai-security-threats-nvidia-nemo-webpage-model-poisoning/</link><guid isPermaLink="true">https://blog.lufsec.com/ai-security-threats-nvidia-nemo-webpage-model-poisoning/</guid><description>A malicious webpage can now silently rewrite the AI model running on your laptop. Here&apos;s what NVIDIA&apos;s NeMo flaw reveals about the career field forming around it.</description><pubDate>Wed, 26 Aug 2026 11:25:47 GMT</pubDate><category>ai security</category><category>Career</category><category>Vulnerabilities</category><category>NVIDIA</category><category>Model Poisoning</category></item><item><title>Poisoning the Watchtower: When Your Logs Become Prompt Injections</title><link>https://blog.lufsec.com/ai-security-threats-prompt-injection-soc-logs-2/</link><guid isPermaLink="true">https://blog.lufsec.com/ai-security-threats-prompt-injection-soc-logs-2/</guid><description>SOCs are feeding attacker-controlled log data to LLM assistants. New research shows how a malicious DNS query can hijack the model and hide the incident.</description><pubDate>Tue, 25 Aug 2026 14:21:58 GMT</pubDate><category>ai security</category><category>prompt injection</category><category>SOC</category><category>LLM Security</category><category>Threat Research</category><category>cybersecurity career</category></item><item><title>IoT Security Vulnerabilities: What the Android Car Malware Story Teaches Beginners</title><link>https://blog.lufsec.com/iot-security-vulnerabilities-android-car-malware-beginner-lessons/</link><guid isPermaLink="true">https://blog.lufsec.com/iot-security-vulnerabilities-android-car-malware-beginner-lessons/</guid><description>Kaspersky&apos;s August 2026 disclosure of malware inside Android car head units is a beginner-friendly window into IoT security vulnerabilities — firmware compromise, poisoned updates, and repurposed devices.</description><pubDate>Mon, 24 Aug 2026 12:48:51 GMT</pubDate><category>iot security</category><category>Vulnerabilities</category><category>Malware</category><category>firmware analysis</category><category>Beginner</category></item><item><title>AI Security Threats: What ATC Research Reveals About Your Career</title><link>https://blog.lufsec.com/ai-security-threats-atc-llm-career-wake-up/</link><guid isPermaLink="true">https://blog.lufsec.com/ai-security-threats-atc-llm-career-wake-up/</guid><description>Researchers put an LLM in the air traffic control loop. If you&apos;re building a cybersecurity career and think AI security is a niche you can defer, this is your wake-up call.</description><pubDate>Fri, 21 Aug 2026 12:28:38 GMT</pubDate><category>ai security</category><category>Career</category><category>LLM</category><category>prompt injection</category><category>Cybersecurity</category></item><item><title>OpenAI&apos;s Training Pause Signals a New Era of AI Security Threats</title><link>https://blog.lufsec.com/openai-training-pause-ai-security-threats-career/</link><guid isPermaLink="true">https://blog.lufsec.com/openai-training-pause-ai-security-threats-career/</guid><description>OpenAI paused frontier RL training for two weeks to shore up defenses. The reason reveals where AI security threats actually live now — and what skills the field will pay for.</description><pubDate>Thu, 20 Aug 2026 12:07:25 GMT</pubDate><category>ai security</category><category>Career</category><category>Cybersecurity</category><category>OpenAI</category><category>Machine Learning Security</category></item><item><title>Unisoc VoLTE Exploit: Why IoT Firmware Analysis Is No Longer Optional</title><link>https://blog.lufsec.com/unisoc-volte-iot-firmware-analysis/</link><guid isPermaLink="true">https://blog.lufsec.com/unisoc-volte-iot-firmware-analysis/</guid><description>A VoLTE video call that ends in full Android kernel access. The Unisoc chain is a case study in why IoT firmware analysis has to reach past the application layer — into the modem&apos;s own OS.</description><pubDate>Wed, 19 Aug 2026 13:38:35 GMT</pubDate><category>iot security</category><category>firmware analysis</category><category>Vulnerability Research</category><category>Mobile Security</category><category>Penetration Testing</category></item><item><title>AI Security Threats: The Adversarial Arms Race Every Practitioner Is Already In</title><link>https://blog.lufsec.com/ai-security-threats-adversarial-arms-race-practitioners/</link><guid isPermaLink="true">https://blog.lufsec.com/ai-security-threats-adversarial-arms-race-practitioners/</guid><description>New research on breaking LLM-powered bot detection is a preview of how adversarial ML will undermine every AI-powered security control. Here&apos;s what practitioners need to understand now.</description><pubDate>Tue, 18 Aug 2026 13:08:55 GMT</pubDate><category>ai security</category><category>Adversarial ML</category><category>cybersecurity career</category><category>Machine Learning Security</category><category>Threat Intelligence</category></item><item><title>IoT Security Vulnerabilities: Why the Radio Layer Is Where Beginners Get Stuck</title><link>https://blog.lufsec.com/iot-security-vulnerabilities-radio-layer-beginners/</link><guid isPermaLink="true">https://blog.lufsec.com/iot-security-vulnerabilities-radio-layer-beginners/</guid><description>Most IoT hacking tutorials stop at firmware. A new open-source tool called TLF proves the real IoT security vulnerabilities live one layer deeper — in the bus traffic between the microcontroller and its radio chip.</description><pubDate>Mon, 17 Aug 2026 13:51:20 GMT</pubDate><category>iot security</category><category>IoT Hacking</category><category>Hardware Security</category><category>RF Security</category><category>Beginner</category></item><item><title>AI Security Threats Just Got Real: What the OpenAI, Anthropic, and Google Reasoning Leak Means for Your Career</title><link>https://blog.lufsec.com/ai-security-threats-openai-anthropic-google-reasoning-leak/</link><guid isPermaLink="true">https://blog.lufsec.com/ai-security-threats-openai-anthropic-google-reasoning-leak/</guid><description>A newly disclosed flaw let researchers replay encrypted reasoning blocks between sessions on OpenAI, Anthropic, and Google APIs — leaking secrets and reshaping what an AI security skill set actually needs to cover.</description><pubDate>Fri, 14 Aug 2026 15:40:19 GMT</pubDate><category>ai security</category><category>Cybersecurity Careers</category><category>LLM Security</category><category>Threat Intelligence</category><category>Pentesting</category></item><item><title>Car Hacking Just Got a New Front Door: The SIM Card Inside Your Telematics Modem</title><link>https://blog.lufsec.com/car-hacking-malicious-sim-card-telematics-modem-attack/</link><guid isPermaLink="true">https://blog.lufsec.com/car-hacking-malicious-sim-card-telematics-modem-attack/</guid><description>Birmingham and Fuzzware researchers showed that a malicious SIM can execute code on the cellular modems inside cars, EV chargers, and industrial IoT. For automotive engineers, that&apos;s a new perimeter — below the stack you already test.</description><pubDate>Thu, 13 Aug 2026 11:13:30 GMT</pubDate><category>car hacking</category><category>automotive security</category><category>iot security</category><category>Cellular Security</category><category>Telematics</category><category>SIM Toolkit</category></item><item><title>Black Hat and DEF CON 34: The Year AI Agents Became the Hacker&apos;s Weapon of Choice</title><link>https://blog.lufsec.com/black-hat-and-def-con-34-the-year-ai-agents-became-the-hackers-weapon-of-choice/</link><guid isPermaLink="true">https://blog.lufsec.com/black-hat-and-def-con-34-the-year-ai-agents-became-the-hackers-weapon-of-choice/</guid><description>AI agents went from novelty to standard hacking weapon this year. My on-the-ground recap of Black Hat and DEF CON 34 — the AI bootcamp, the villages, 16 interviews, and the launch of Inside Cyber Minds Season 3.</description><pubDate>Mon, 10 Aug 2026 13:19:40 GMT</pubDate></item><item><title>IoT Firmware Analysis: What TP-Link&apos;s Omada ZTP Patch Reveals About the Attack Surface You&apos;re Missing</title><link>https://blog.lufsec.com/iot-firmware-analysis-tp-link-omada-ztp-flaws/</link><guid isPermaLink="true">https://blog.lufsec.com/iot-firmware-analysis-tp-link-omada-ztp-flaws/</guid><description>TP-Link just patched 15 zero-touch provisioning flaws in Omada gear. The real story isn&apos;t the CVEs — it&apos;s the firmware attack surface most IoT pen testers never open.</description><pubDate>Wed, 05 Aug 2026 14:28:06 GMT</pubDate><category>iot security</category><category>firmware analysis</category><category>Vulnerability Research</category><category>Pen Testing</category><category>TP-Link</category></item><item><title>AI Security Threats Are Just Old Security Failures in New Packaging</title><link>https://blog.lufsec.com/ai-security-threats-anthropic-claude-iam-fundamentals-career/</link><guid isPermaLink="true">https://blog.lufsec.com/ai-security-threats-anthropic-claude-iam-fundamentals-career/</guid><description>Anthropic&apos;s post-mortem on last month&apos;s Claude incidents blames over-permissioning, not the model. For anyone entering cybersecurity, that&apos;s the most important career signal of the year.</description><pubDate>Tue, 04 Aug 2026 15:40:05 GMT</pubDate><category>ai security</category><category>Career</category><category>IAM</category><category>Least Privilege</category><category>Cybersecurity Fundamentals</category></item><item><title>Claude Went Rogue: What Anthropic&apos;s Breach Tests Reveal About AI Security Threats</title><link>https://blog.lufsec.com/claude-anthropic-pypi-breach-ai-security-threats/</link><guid isPermaLink="true">https://blog.lufsec.com/claude-anthropic-pypi-breach-ai-security-threats/</guid><description>Anthropic&apos;s Claude autonomously breached three organizations and pushed malware to PyPI during safety evaluations. The lesson for defenders — and for anyone entering security — isn&apos;t about jailbreaks.</description><pubDate>Fri, 31 Jul 2026 13:49:13 GMT</pubDate><category>ai security</category><category>Threat Intelligence</category><category>Supply Chain Security</category><category>Cybersecurity Careers</category><category>anthropic</category></item><item><title>Prompt Injection Is the Attack Nobody Trained For. So I Built the Course.</title><link>https://blog.lufsec.com/prompt-injection-course-launch/</link><guid isPermaLink="true">https://blog.lufsec.com/prompt-injection-course-launch/</guid><description>Companies are wiring LLMs into real systems without understanding prompt injection. I built a hands-on course that teaches the attack, and the defense. 2 free lessons, no card.</description><pubDate>Fri, 17 Jul 2026 15:46:58 GMT</pubDate></item><item><title>AI Security Threats: When Attackers Talk Directly to Your SOC&apos;s LLM</title><link>https://blog.lufsec.com/ai-security-threats-prompt-injection-llm-soc-logs/</link><guid isPermaLink="true">https://blog.lufsec.com/ai-security-threats-prompt-injection-llm-soc-logs/</guid><description>New research shows attackers can hide prompt injection payloads inside routine log fields — turning LLM-augmented SOC tools into unwitting accomplices. Here&apos;s what that means for anyone building a security career.</description><pubDate>Wed, 15 Jul 2026 21:02:54 GMT</pubDate><category>ai security</category><category>prompt injection</category><category>SOC</category><category>LLM Security</category><category>Cybersecurity Careers</category><category>Threat Research</category></item><item><title>Poisoning the Watchtower: Prompt Injection Turns Your SOC&apos;s AI Against You</title><link>https://blog.lufsec.com/ai-security-threats-prompt-injection-soc-log-poisoning/</link><guid isPermaLink="true">https://blog.lufsec.com/ai-security-threats-prompt-injection-soc-log-poisoning/</guid><description>Attackers can now smuggle instructions into log fields your SOC&apos;s AI assistant reads. Here&apos;s how the watchtower gets poisoned — and why every career-minded security pro needs to understand it.</description><pubDate>Tue, 14 Jul 2026 12:56:27 GMT</pubDate><category>ai security</category><category>prompt injection</category><category>SOC</category><category>LLM Security</category><category>cybersecurity career</category><category>Threat Intelligence</category></item><item><title>IoT Security Vulnerabilities Start at Boot: What the U-Boot Flaws Teach Beginners</title><link>https://blog.lufsec.com/iot-security-vulnerabilities-u-boot-flaws-beginners/</link><guid isPermaLink="true">https://blog.lufsec.com/iot-security-vulnerabilities-u-boot-flaws-beginners/</guid><description>Six new U-Boot bootloader flaws show why IoT security vulnerabilities start before the OS loads — and why beginners should learn the boot chain first.</description><pubDate>Mon, 13 Jul 2026 17:26:45 GMT</pubDate><category>iot security</category><category>Firmware</category><category>Bootloader</category><category>U-Boot</category><category>Beginner</category><category>Vulnerabilities</category></item><item><title>AI Security Threats: Why the &apos;Friendly Fire&apos; Attack Makes Your Cybersecurity Career More Valuable, Not Less</title><link>https://blog.lufsec.com/ai-security-threats-friendly-fire-cybersecurity-career/</link><guid isPermaLink="true">https://blog.lufsec.com/ai-security-threats-friendly-fire-cybersecurity-career/</guid><description>AI agents built to catch malicious code can be tricked into running it. Here&apos;s why that paradox makes human security expertise more essential than ever.</description><pubDate>Fri, 10 Jul 2026 17:30:28 GMT</pubDate><category>ai security</category><category>cybersecurity career</category><category>prompt injection</category><category>Adversarial AI</category><category>Threat Intelligence</category></item><item><title>Car Hacking: ECUInjector and the Automotive Offensive Skill Gap</title><link>https://blog.lufsec.com/car-hacking-ecuinjector-automotive-offensive-skill-gap/</link><guid isPermaLink="true">https://blog.lufsec.com/car-hacking-ecuinjector-automotive-offensive-skill-gap/</guid><description>A new open-source ECU security analysis tool, ECUInjector, exposes the car hacking tradecraft most automotive engineers never learn — and why it&apos;s now table stakes for building defensible vehicles.</description><pubDate>Thu, 09 Jul 2026 19:54:01 GMT</pubDate><category>car hacking</category><category>automotive security</category><category>ECU</category><category>Embedded Security</category><category>offensive security</category></item><item><title>AI Security Threats and the Trust Surfaces Attackers Keep Exploiting</title><link>https://blog.lufsec.com/ai-security-threats-trust-surfaces-weekly-recap/</link><guid isPermaLink="true">https://blog.lufsec.com/ai-security-threats-trust-surfaces-weekly-recap/</guid><description>This week&apos;s incidents — proxy botnets, browser ransomware, AI agent prompt injection, malicious PoC repos — share one pattern: misplaced trust in ordinary components. Here&apos;s how to see it and audit it.</description><pubDate>Tue, 07 Jul 2026 22:23:04 GMT</pubDate><category>ai security</category><category>Threat Intelligence</category><category>Career</category><category>Trust Models</category><category>Weekly Recap</category></item><item><title>AI Security Threats Are Moving Into the Model: What kNNGuard Signals for Your Career</title><link>https://blog.lufsec.com/ai-security-threats-knnguard-activations-career/</link><guid isPermaLink="true">https://blog.lufsec.com/ai-security-threats-knnguard-activations-career/</guid><description>A new training-free guardrail called kNNGuard reads an LLM&apos;s hidden activations to catch adversarial prompts. Here&apos;s why that shifts the AI security skill floor — and what employers will expect from the next hiring tier.</description><pubDate>Fri, 03 Jul 2026 14:11:03 GMT</pubDate><category>ai security</category><category>Career</category><category>LLM</category><category>Guardrails</category><category>Threat Research</category></item><item><title>Car Hacking: LIN-Bus Fuzzing Is the Attack Surface You&apos;ve Been Ignoring</title><link>https://blog.lufsec.com/lin-bus-fuzzing-car-hacking-bb-fast-uds/</link><guid isPermaLink="true">https://blog.lufsec.com/lin-bus-fuzzing-car-hacking-bb-fast-uds/</guid><description>Your CAN fuzzer doesn&apos;t speak LIN — and the LIN bus now carries UDS diagnostics on door locks, mirrors, and gateway ECUs. A 2026 paper on BB-FAST shows how Bayesian pruning cracks the bandwidth-starved bus most car hackers ignore.</description><pubDate>Thu, 02 Jul 2026 14:14:17 GMT</pubDate><category>car hacking</category><category>automotive security</category><category>LIN bus</category><category>UDS</category><category>fuzzing</category><category>Embedded Security</category></item><item><title>libssh2 CVE-2026-55200 Turns Your IoT Firmware Into the Attack Surface</title><link>https://blog.lufsec.com/libssh2-cve-55200-iot-firmware-analysis-wake-up-call/</link><guid isPermaLink="true">https://blog.lufsec.com/libssh2-cve-55200-iot-firmware-analysis-wake-up-call/</guid><description>A new libssh2 client-side flaw flips the IoT threat model. Here&apos;s how to tear open firmware and find the vulnerable libraries hiding inside.</description><pubDate>Wed, 01 Jul 2026 15:24:07 GMT</pubDate><category>iot security</category><category>firmware analysis</category><category>CVE</category><category>SSH</category><category>Pen Testing</category><category>libssh2</category></item><item><title>AI Security Threats: Why LLM Code Scanners Inherit Human Bias</title><link>https://blog.lufsec.com/ai-security-threats-llm-cognitive-heuristics-code-vulnerability-detection/</link><guid isPermaLink="true">https://blog.lufsec.com/ai-security-threats-llm-cognitive-heuristics-code-vulnerability-detection/</guid><description>New research shows LLM vulnerability scanners fall for the same cognitive shortcuts that fool human analysts. Here&apos;s what that means if you&apos;re building a security career around AI tools.</description><pubDate>Tue, 30 Jun 2026 11:46:29 GMT</pubDate><category>ai security</category><category>Career</category><category>Vulnerability Detection</category><category>LLM</category><category>Code Review</category></item><item><title>IoT Security Vulnerabilities Hiding in Clock Sync: The UWB Timing Attack Surface Nobody Teaches</title><link>https://blog.lufsec.com/iot-security-vulnerabilities-uwb-clock-sync-attack-surface/</link><guid isPermaLink="true">https://blog.lufsec.com/iot-security-vulnerabilities-uwb-clock-sync-attack-surface/</guid><description>IoT security vulnerabilities live in the timing layer too. UWB location networks depend on nanosecond clock sync between anchors — and that sync layer is its own attack surface.</description><pubDate>Mon, 29 Jun 2026 12:31:19 GMT</pubDate><category>iot security</category><category>UWB</category><category>Wireless Attacks</category><category>Industrial IoT</category><category>SDR</category><category>Beginner</category></item><item><title>AI Security Threats: What the Résumé Prompt Injection Research Means for Your Career</title><link>https://blog.lufsec.com/ai-security-threats-resume-prompt-injection-career/</link><guid isPermaLink="true">https://blog.lufsec.com/ai-security-threats-resume-prompt-injection-career/</guid><description>Researchers showed candidates can prompt-inject LLM résumé screeners to game hiring. Here&apos;s what that says about AI security threats — and the career edge it hands you.</description><pubDate>Fri, 26 Jun 2026 14:40:36 GMT</pubDate><category>ai security</category><category>prompt injection</category><category>cybersecurity career</category><category>Adversarial ML</category><category>LLM Security</category></item><item><title>Poisoning the Watchtower: How Prompt Injection Turns Your AI SOC Analyst Against You</title><link>https://blog.lufsec.com/ai-security-threats-prompt-injection-soc-logs/</link><guid isPermaLink="true">https://blog.lufsec.com/ai-security-threats-prompt-injection-soc-logs/</guid><description>New research shows attackers can hijack the logs an AI SOC analyst reads, manipulating triage decisions and suppressing alerts. Here&apos;s what that means for your career.</description><pubDate>Thu, 25 Jun 2026 13:58:40 GMT</pubDate><category>ai security</category><category>SOC</category><category>prompt injection</category><category>Career</category><category>Threat Research</category></item><item><title>IoT Firmware Analysis: What GitHub&apos;s Pwn Request Patch Teaches Embedded Hunters</title><link>https://blog.lufsec.com/iot-firmware-analysis-github-pwn-request-trust-boundaries/</link><guid isPermaLink="true">https://blog.lufsec.com/iot-firmware-analysis-github-pwn-request-trust-boundaries/</guid><description>GitHub just hardened actions/checkout against pwn request attacks. The same trust-boundary failure sits in countless IoT devices — here&apos;s how to find it during firmware analysis.</description><pubDate>Wed, 24 Jun 2026 11:20:34 GMT</pubDate><category>iot security</category><category>firmware analysis</category><category>reverse engineering</category><category>Supply Chain Security</category><category>Ghidra</category></item><item><title>AI Security Threats Are Reshaping Entry-Level Cybersecurity Jobs</title><link>https://blog.lufsec.com/ai-security-threats-gpt-55-cyber-career/</link><guid isPermaLink="true">https://blog.lufsec.com/ai-security-threats-gpt-55-cyber-career/</guid><description>OpenAI&apos;s GPT-5.5-Cyber automates vulnerability research at scale. Here&apos;s what that actually means for anyone trying to break into cybersecurity right now.</description><pubDate>Tue, 23 Jun 2026 12:19:46 GMT</pubDate><category>ai security</category><category>cybersecurity career</category><category>Vulnerability Research</category><category>GPT-5.5-Cyber</category><category>OpenAI Daybreak</category></item><item><title>Canada&apos;s Spy Agency Hacked Your Router. Here&apos;s What That Says About IoT Security Vulnerabilities.</title><link>https://blog.lufsec.com/csis-botnet-warrant-iot-security-vulnerabilities/</link><guid isPermaLink="true">https://blog.lufsec.com/csis-botnet-warrant-iot-security-vulnerabilities/</guid><description>Canada&apos;s intelligence service got a court warrant to break into infected home routers and IoT gear on Canadian soil. The vulnerabilities they exploited are the same ones beginners can learn to find.</description><pubDate>Mon, 22 Jun 2026 13:14:34 GMT</pubDate><category>iot security</category><category>botnets</category><category>Vulnerabilities</category><category>Beginner</category><category>News Analysis</category></item><item><title>AI Security Threats: Why Multi-Turn Jailbreaks Change the Career Game</title><link>https://blog.lufsec.com/ai-security-threats-multi-turn-jailbreaks-career/</link><guid isPermaLink="true">https://blog.lufsec.com/ai-security-threats-multi-turn-jailbreaks-career/</guid><description>A new arXiv benchmark jailbreaks LLM agents running a simulated nuclear plant — over many turns, not one prompt. Here&apos;s what that means for anyone building an AI security career.</description><pubDate>Fri, 19 Jun 2026 12:40:36 GMT</pubDate><category>ai security</category><category>Career</category><category>Red Teaming</category><category>LLM Security</category><category>Threat Modeling</category></item></channel></rss>