From MCP Risks to AI Jailbreaks with Marco Figueroa

When AI agents move faster than security teams, the game changes, and the risks multiply. Ron welcomes back Marco “Mystic Marc” Figueroa, Program Manager at Mozilla’s 0DIN Program, to continue the conversation and update on 2025’s most pressing AI and cybersecurity shifts. From the explosive rise of AI agents and OpenAI’s rumored browser to the hidden dangers of MCP implementations and prompt injection exploits like the Gemini attack, Marco shares insights that security pros can’t afford to miss. Impactful Moments 00:00 - Introduction 02:00 - Why 2025 is the year of the agent 05:45 - MCP’s rapid adoption and security risks 10:00 - The Gemini prompt injection vulnerability 15:00 - How attackers hide malicious AI prompts 18:00 - High success rates in non-technical teams 22:00 - Rise of voice-based AI scams 25:00 - Using jailbreaks to bend AI to your needs 30:00 - Predictions on OpenAI’s upcoming browser 33:00 - The profit battle between OpenAI and Microsoft 35:00 - Windsurf’s rollercoaster of acquisitions Links: Connect with our guest Marco on LinkedIn: https://www.linkedin.com/in/marco-figueroa-re/ Check out our upcoming events: https://www.hackervalley.com/livestreams Join our creative mastermind and stand out as a cybersecurity professional: https://www.patreon.com/hackervalleystudio Love Hacker Valley Studio? Pick up some swag: https://store.hackervalley.com Continue the conversation by joining our Discord: https://hackervalley.com/discord Become a sponsor of the show to amplify your brand: https://hackervalley.com/work-with-us/