19 articles with this tag
The CTF Escape Horizon: AI Agent Model Breakouts Explained
An analysis of the July 2026 OpenAI and Anthropic sandbox escapes, zero-day Artifactory exploits, misconfigured evals, and microVM isolation.
The Model Context Collapse: Why AI Coding Agents Forget
AI coding agents fail silently in long sessions due to context degradation. We analyze the architecture of 'context rot' and how context caching solves it.
The Vibe & Verify Fallacy: Why AI-Generated Tests Are Creating a False Sense of Code Quality
With 84% of devs using AI daily, 'vibe and verify' is the new workflow. But letting AI write tests for its own code creates a confirmation bias trap.
The Alignment Tax: ASI09 & ASI10 — Your Agent IS the Threat
OWASP Agentic Top 10 series finale. ASI09 (Trust Exploitation) and ASI10 (Rogue Agents) — the threats that don't need an external attacker.
The Trivy Cascade: 75 Poisoned Tags, a Blockchain Worm, 5 Days of Chaos
A complete technical dissection of the TeamPCP supply chain attack that cascaded from Trivy to Checkmarx to npm to PyPI — the largest CI/CD compromise of 2026.
BitTorrent's Creator Says Git Is Broken — 470 Lines of Python Prove It
Bram Cohen's Manyana uses CRDTs so merges never fail. With Jujutsu at 27K stars and agents making thousands of commits, Git's merge model is under siege.
NVIDIA NemoClaw: The SELinux for Agent Governance
NVIDIA NemoClaw adds kernel-level sandboxing and out-of-process policy enforcement to OpenClaw. We map its architecture against the OWASP Agentic Top 10.
When One Agent Falls, They All Fall: ASI07 & ASI08 — The Distributed Systems Nightmare That Multi-Agent Architectures Weren't Built to Survive
Amazon's Kiro deleted a production environment. 40% of multi-agent pilots fail in 6 months. ASI07 and ASI08 are distributed systems problems in AI costumes.
Frameworks Aren't Dead. They're the Reason Your Agent Can Write Code at All.
83% React usage, 20M weekly downloads, and AI agents that can't function without frameworks. Icarus declared them dead. The data says otherwise.
The Invisible Wire: 175,000 Naked AI Agents, a WireGuard Mesh, and Why Tailscale Is Becoming the Nervous System of Agentic Infrastructure
175,000 AI agents exposed with zero authentication. The network layer nobody is securing, and how a WireGuard mesh with identity-based access closes it.
When Your Agent Becomes the Exploit: ASI05 & ASI06 — The Twin Threats That Turn AI Autonomy Against You
A deep dive into OWASP Agentic ASI05 and ASI06: the Claude Code CVEs, the Summer Yue incident, and how code execution chains with memory poisoning.
Frameworks Are Dead. Architects Are Not.
57% of companies run AI agents in production. 600 HN comments on one post. The framework era is ending — here's what replaces it.
The OpenClaw Meltdown: 9 CVEs, 2,200 Malicious Skills, and the Most Comprehensive Real-World Test of the OWASP Agentic Top 10
In 5 weeks, OpenClaw triggered 8 of 10 OWASP Agentic vulnerability classes. This is the forensic dissection every engineer building AI agents needs to read.
GPT-5.2 Just Solved a 15-Year Physics Mystery — Then Scored 0% on the Physics Exam
OpenAI's GPT-5.2 derived a new formula for gluon amplitudes that eluded top physicists for 15 years, then scored 0% on CritPt.
The New Security Bible: Why Every Engineer Building AI Agents Needs the OWASP Agentic Top 10
The OWASP Agentic Top 10 is the first security framework for autonomous AI agents. 10 risks, real incidents, and the attack chains that connect them.
The Week Software Broke: $285B Wiped, Two AI Superpowers Collide, and the SaaS Model Starts to Crack
In one week, Anthropic's Cowork plugins crashed $285B in software stocks, then both Opus 4.6 and GPT-5.3 Codex launched within 27 minutes.
FREE FOR 7 DAYS ONLY: Kilocode Unleashed Kimi K2.5 and Put Cursor on Life Support
🔥 FREE until Feb 3! Kilocode + Kimi K2.5 (1T params) vs Cursor: $0 unlimited AI coding. Agent Swarm, Vision, 256K context. Install in 30 seconds.
The Code That Changed Everything: How to Build a Moltbook Agent That Actually Works
A veteran's guide to production-grade Moltbook agents: security sandboxing, agent-to-agent protocols, and the failure modes tutorials never mention.
The Syntax Singularity: How One Developer Built a Programming Language in 24 Hours
Bernard Lambeau created Elo, a cross-compiling language targeting JS, Ruby, and SQL, in 24 hours using Claude Code. The Man-Month is dead.