Subvolts take Secure AI agents by understanding and testing for new exploitation risks. Oct 16, 2025 - Developers
- Learn about vulnerabilities in AI agents and how to use tools like Spikee to test and secure them.
- Business
- Understand the emerging security risks associated with deploying complex AI agents in your operations.
This episode offers a solid interview with an expert on AI agent security, discussing practical exploitation and defense strategies.
Dealing with increasingly complicated agents
Audio streams from Daniel Whitenack and Chris Benson. Your place is remembered on this device.
From Subvolts
Summary and analysis by Subvolts editorial, AI-assisted.
In this Practical AI episode, Daniel Whitenack and Chris Benson interview Donato Capitella about security risks in complex AI agentic workflows. Capitella explains how intricate architectures make agents fragile and vulnerable to prompt injection, data exfiltration, and tool misuse. He shares real-world penetration test stories and discusses his open-source toolkit, Spikee, for evaluating AI system security.
AI systems are becoming more complex, like intricate machines with many parts. This complexity can create hidden weaknesses that attackers can exploit, similar to finding a loose screw. This episode explains these risks and how to test for them, so AI systems can be made safer.
This episode is for AI developers and security professionals concerned with the growing complexity of AI agents. Capitella's insights into prompt injection, data exfiltration, and tool misuse directly address emerging security challenges as AI systems become more sophisticated. Listeners learn about practical exploitation methods and how tools like Spikee can help build more robust and secure AI agents, relevant to ongoing efforts by labs like OpenAI and Google Gemini to enhance AI safety and reliability.
- HostDaniel Whitenack
- HostChris Benson
- GuestDonato Capitella
- Episode Title: Dealing with increasingly complicated agents
- Show: Practical AI
- Guest: Donato Capitella
- Length: 55 minutes
- Security Risks: Prompt injection, data exfiltration, and tool misuse
- Toolkit Mentioned: Spikee (Simple Prompt Injection Kit for Evaluation and Exploitation)
Chapters from the publisher ↗ · Transcript ↗ Key moments and transcript search come with a later piece.
As AI systems move from simple chatbots to complex agentic workflows, new security risks emerge. In this episode, Donato Capitella unpacks how increasingly complicated architectures are making agents fragile and vulnerable. These agents can be exploited through prompt injection, data exfiltration, and tool misuse. Donato shares stories from real-world penetration tests, the design patterns for building LLM agents and explains how his open-source toolkit Spikee (Simple Prompt Injection Kit for Evaluation and Exploitation) is helping red teams probe AI systems.
Featuring:
- Donato Capitella – LinkedIn, X
- Chris Benson – Website, LinkedIn, Bluesky, GitHub, X
- Daniel Whitenack – Website, GitHub, X
Links:
Sponsors:
- Outshift by Cisco - The open source collective building the Internet of Agents. Backed by Outshift by Cisco, AGNTCY gives developers the tools to build and deploy multi-agent software at scale. Identity, communication protocols, and modular workflows—all in one global collaboration layer. Start building at AGNTCY.org.
- Shopify – The commerce platform trusted by millions. From idea to checkout, Shopify gives you everything you need to launch and scale your business—no matter your level of experience. Build beautiful storefronts, market with built-in AI tools, and tap into the platform powering 10% of all U.S. eCommerce. Start your one-dollar trial at shopify.com/practicalai
- Fabi.ai - The all-in-one data analysis platform for modern teams. From ad hoc queries to advanced analytics, Fabi lets you explore data wherever it lives—spreadsheets, Postgres, Snowflake, Airtable and more. Built-in Python and AI assistance help you move fast, then publish interactive dashboards or automate insights delivered straight to Slack, email, spreadsheets or wherever you need to share it. Learn more and get started for free at fabi.ai
Upcoming Events:
- Join us at the Midwest AI Summit on November 13 in Indianapolis to hear world-class speakers share how they’ve scaled AI solutions. Don’t miss the AI Engineering Lounge, where you can sit down with experts for hands-on guidance. Reserve your spot today!
- Register for upcoming webinars here!
- agentic workflows
- Donato Capitella
- AI systems that can perform tasks and make decisions autonomously.
- prompt injection
- Donato Capitella
- Tricking an AI by embedding hidden instructions within its input.
- data exfiltration
- Donato Capitella
- Unauthorized transfer of sensitive data from an AI system.
44 minNarrative intelligence and the human advantage
WithReed Frerichs
47 minOpen models and the future of Physical AI with NVIDIA
WithMing-Yu Liu
49 minFrom AGENTS.md to Enterprise Deployment
WithNick Kuhn
1 h 03Samsung Galaxy XR Gets Real
WithTom Merritt
1 h 18Why GM will give you Gemini — but not CarPlay
WithMary Barra, Sterling Anderson
1 hGoogle Researcher Shows Life "Emerges From Code" - Blaise Agüera y Arcas
WithBlaise Agüera y Arcas
44 minQuestions people ask
- What are the main security risks with complex AI agents?
- Complex AI agents face risks like prompt injection, data exfiltration, and tool misuse due to their intricate architectures.
- What is Spikee?
- Spikee is an open-source toolkit designed to help red teams evaluate and exploit AI systems, particularly for prompt injection vulnerabilities.
Page generated Oct 19, 2026. Summaries are Subvolts' own; the audio, artwork and notes belong to Daniel Whitenack and Chris Benson.