Sign up to save your podcasts
Or
Share The Dark Side of Agentic AI: 322% More Vulnerabilities & The End of Junior Devs?
Share to email
Share to Facebook
Share to X
Share to email
Share to Facebook
Share to X
Or
The Dark Side of Agentic AI: 322% More Vulnerabilities & The End of Junior Devs?
The era of AI as a simple autocomplete tool is officially over. Welcome to the age of "Agentic Development", but are we trading development speed for catastrophic security failures?
In this episode of The Debates, we unpack the massive shift from basic coding assistants to autonomous AI agents, like Google Anti-Gravity and Claude Code, which plan, execute, and verify tasks entirely on their own. While moving from a code editor to a "mission control center" sounds incredible, the rise of "vibe coding", generating code based on how it feels without understanding the underlying logic is creating a terrifying reality.
We dive deep into the hard data and explore why blind trust in AI is a recipe for disaster.
In this episode, we uncover:
Why AI-generated code currently contains 2.74 times more vulnerabilities than human-written code.
How security firm APIro discovered a staggering 322% increase in privilege escalation paths and a 153% increase in design flaws in AI-assisted projects.
The alarming failure rate of AI code in specific languages, including a 72% failure rate for Java secure coding benchmarks.
The "Context Rot" problem, which explains how large language models forget crucial security constraints as your project expands.
The Pyramid Replacement crisis: if AI handles all entry-level work, how will junior developers ever gain the expertise needed to audit complex AI outputs?.
How to fight back using the Software Ceni framework and by applying a strict "Zero Trust" model to your AI agents.
Stop committing code you don't understand. Tune in to learn how to survive the shift to Agentic AI by becoming a true system architect.
🎧 Listen now before you merge that next AI pull request!
#SoftwareEngineering #AgenticAI #CyberSecurity #TechPodcast #VibeCoding #FutureOfWork #Developers
View all episodes
By Revedor AIThe era of AI as a simple autocomplete tool is officially over. Welcome to the age of "Agentic Development", but are we trading development speed for catastrophic security failures?
In this episode of The Debates, we unpack the massive shift from basic coding assistants to autonomous AI agents, like Google Anti-Gravity and Claude Code, which plan, execute, and verify tasks entirely on their own. While moving from a code editor to a "mission control center" sounds incredible, the rise of "vibe coding", generating code based on how it feels without understanding the underlying logic is creating a terrifying reality.
We dive deep into the hard data and explore why blind trust in AI is a recipe for disaster.
In this episode, we uncover:
Why AI-generated code currently contains 2.74 times more vulnerabilities than human-written code.
How security firm APIro discovered a staggering 322% increase in privilege escalation paths and a 153% increase in design flaws in AI-assisted projects.
The alarming failure rate of AI code in specific languages, including a 72% failure rate for Java secure coding benchmarks.
The "Context Rot" problem, which explains how large language models forget crucial security constraints as your project expands.
The Pyramid Replacement crisis: if AI handles all entry-level work, how will junior developers ever gain the expertise needed to audit complex AI outputs?.
How to fight back using the Software Ceni framework and by applying a strict "Zero Trust" model to your AI agents.
Stop committing code you don't understand. Tune in to learn how to survive the shift to Agentic AI by becoming a true system architect.
🎧 Listen now before you merge that next AI pull request!
#SoftwareEngineering #AgenticAI #CyberSecurity #TechPodcast #VibeCoding #FutureOfWork #Developers