![\"About](\"https:\/\/blog.kybervandals.com\/content\/images\/2026\/02\/photo-1471724603960-98bc193c621d.jpeg\")
<\/p>\n\n<\/p>\n\n
Oh Boy, Oh Boy, what crazy times we live in.<\/p>\n
12 Months ago the fuzz was all about MCP, then we saw the rise of agentic AI and skills in a short timeframe, and now its all about agentic swarms, OpenClaw, personal agents etc.<\/p>\n
In June 2025 we published our first article „No More Secrets: The Use of Offensive AI<\/a>“ which took a snapshot of what was the state of AI\/LLM in offensive security at that time, and our conclusion was: <\/p>\n At that moment, AI was supporting different stages of an attack without the ability to go beyond script kiddy level and execute complex tasks like recon, choosing the right attack paths, adjusting playloads, beeing stealthy as possible etc. <\/p>\n Now, just 9 months later, the landscape has changed drastically, with no end in sight. <\/p>\n Let me give some examples from just 2026 alone that shows the path AI in offensive security heads to:<\/p>\n the cybersecurity industry is about to get completely disrupted.. \ud83d\udc80<\/p>\n someone just open-sourced a fully autonomous AI Red Team.<\/p>\n it\u2019s called PentAGI. multiple AI agents that talk to each other to hack a target. zero human input.pic.twitter.com\/NQdyTs1LGw<\/a><\/p>\n \u2014 Oliver Prompts (@oliviscusAI) February 19, 2026<\/a><\/p><\/blockquote>\n<\/figure>\n In mid-September 2025, we detected suspicious activity that later investigation determined to be a highly sophisticated espionage campaign. The attackers used AI\u2019s \u201cagentic\u201d capabilities to an unprecedented degree\u2014using AI not just as an advisor, but to execute the cyberattacks themselves.<\/p><\/blockquote>\n AI models can now find high-severity vulnerabilities at scale. Our view is this is a moment to move quickly\u2014to empower defenders and secure as much code as possible while the window exists.<\/p><\/blockquote>\n\nAI makes the good ones faster, attacks cheaper and the impact more precise. Also, AI is currently implemented into attack-tools, which means no attackvectors will remain hidden, leading to an sharp increase of successrates.<\/strong><\/b><\/div>\n<\/div>\n
\n
\n<\/p>\nPentAGI, an autonomous RedTeaming Agent<\/h3>\n
\n
\n
Anthropic disrupts AI espionage<\/h3>\n\n
\n
0-Day-Flood by AI expected<\/h3>\n\n