Ep. 23 - Knight in shinning PromptArmor with Vikram Jayanthi | Developers:After Dark

Is your cutting-edge AI also your biggest vulnerability? Dive deep into the world of AI Security as we uncover how seemingly harmless data can become an attacker's instructions. We're joined by Vikram, co-founder and CTO of Prompt Armor, to explore the new frontier of AI Risk Management and why traditional security measures might not be enough.In this eye-opening episode, Vikram demystifies the real threats lurking in Large Language Models (LLMs), from the nuances ofPrompt Injection to the dangers of malicious data in your systems. Discover the critical differences between conventional cybersecurity and the unique challenges posed by AI, including how text itself can become malware. Vikram also sheds light on common misconceptions about AI security, the role of emerging standards like the OWASP Top 10 for LLM, and what skills aspiring AI security professionals need to cultivate. Learn practical steps companies can take when deploying AI internally and gain insights into the evolving global landscape of AI adoption and security.Chapters:(00:00:00) Intro & Welcome Vikram from Prompt Armor(00:01:54) What is Prompt Armor?(00:03:30) The Genesis of Prompt Armor: Seeing the AI Security Gap Early(00:06:30) AI Security: Cutting Through the Hype and Fear(00:08:09) Learning AI Security: Prompt Armor's Free Course(00:11:50) Traditional vs. AI Security: Why "Data as Instruction" Changes Everything(00:13:50) The Threat of Indirect Prompt Injection & MXGp Servers(00:20:15) Misconceptions Organizations Have About AI Security(00:22:30) Government Regulation vs. Industry Innovation in AI Security(00:24:50) The Evolving AI Security Landscape: What's Next?(00:26:15) Top 3 Steps for Companies Deploying AI Internally(00:31:00) Skills for Aspiring AI Security Professionals(00:34:00) Is it Really AI? Distinguishing Hype from Genuine AI Systems(00:46:40) Common Myths About Working in AI Security(00:49:00) AI, Voice, and Video: The Next Security Challenges?(01:00:45) Vikram's Favorite Part of Working in AI Security & Plugs#AISecurity #Cybersecurity #LLMSecurity #TechPodcast #PromptInjectionResources Mentioned:Prompt Armor: https://promptarmor.comPrompt Armor's Free AI Security Course (via Third Party Risk Association): https://www.linkedin.com/posts/tpra_promptarmor-activity-7333114864887865346-GWNUOWASP Top 10 for LLM Applications:https://owasp.org/www-project-top-10-for-large-language-model-applications/We aim to incorporate the human element and drive a conversation that is fun, informational, and thought-provoking. You can watch Developers After Dark on the HubSpot developer’s channel or wherever you get your podcasts.Check out the Developers:after Dark website atdevsafterdarkpod.comWant to be a guest?Fill out the form 👉https://share.hsforms.com/2J5me9_v3Rlix21-1LoZWkAbpd9r