ChAI Chat

ChAI Chat Episode 4: Autonomous Vehicle Safety with Mateo Delgado

In this episode, Jomar Gacoscos and guest Matteo Delgado delve into the complexities of autonomous vehicles, discussing their ethical implications, safety challenges, and the importance of infrastructure. They explore the various levels of autonomy, the role of cybersecurity, and the need for collaboration between companies and city planners to ensure safe and effective deployment of this technology. The conversation highlights the current state of autonomous driving and the hurdles that still need to be overcome for widespread adoption. Disclaimer: The views and opinions expressed in this podcast are solely those of the host and guest and do not necessarily reflect the official policies or positions of our respective employers or affiliated organizations. The content is intended for informational and entertainment purposes only and should not be construed as professional advice. Additional Resources https://www.moralmachine.net/ [https://www.moralmachine.net/] https://www.media.mit.edu/projects/moral-machine/overview/ [https://www.media.mit.edu/projects/moral-machine/overview/] Phil Koopman - How Safe is Safe Enough, Measuring and Predicting Autonomous Vehicle Safety Koopman P, Wagner M. - Autonomous vehicle safety: An interdisciplinary challenge. Koopman P. - Challenges in Autonomous Vehicle Testing and Validation Autonomous Vehicle Safety: Lessons from Aviation By Jaynarayan H. Lala, Carl E. Landwehr, and John F. Meyer NHTSA (National Highway Traffic Safety Administration) - Autonomous Vehicle Safety Reports and Audits Chapters 00:00 Introduction to Autonomous Vehicles 03:40 Understanding Autonomous Vehicle Levels 11:55 Defining Safety in Autonomous Driving 19:07 Operational Design Domains and Environmental Challenges 27:06 Training Autonomous Vehicles for Diverse Environments 33:03 Cybersecurity Challenges in Autonomous Vehicles 38:37 Cybersecurity Controls in Autonomous Vehicles 46:52 Trust and Control in Autonomous Systems 57:52 Collaboration Between Companies and Infrastructure 01:08:17 The Future of Autonomous Vehicles and Road Safety

ChAI Chat Episode 3: Voice Cloning and AI Safety

In this episode of the Chai Chat podcast, Jomar Gacoscos explores the fascinating world of AI voice cloning and synthetic voice technology. He discusses its applications in gaming and media, the potential risks associated with voice theft and identity fraud, and the importance of AI safety. The conversation highlights the rapid evolution of voice cloning technology and the need for preventive measures to mitigate its misuse. Chapters 00:00 Introduction to AI Voice Cloning 03:51 The Dark Side of Voice Cloning 06:22 How Eleven Labs is Trying to Prevent Voice Theft and Ensuring Safety 08:03 Conclusion and Reflection on AI Voice Technology Show Notes * https://elevenlabs.io/ [https://elevenlabs.io/] * Sony Music says it has removed over 75,000 deepfake tracks from streaming platforms, which include voice models of Beyoncé, Harry Styles and Queen [https://www.musicradar.com/music-industry/sony-music-says-it-has-removed-over-75-000-deepfake-tracks-from-streaming-platforms-which-include-voice-models-of-beyonce-harry-styles-and-queen]

ChAI Chat Episode 2: AI Risk - An Overiew

In the second episode of ChAI Chat, host Jomar Gacoscos, an information security professional, explores AI risks and safety concerns in an era of rapid technological advancement. He discusses how AI can be manipulated, citing examples like the "Do Anything Now" (DAN) prompt, which bypassed ChatGPT’s safeguards, and a Chevrolet dealership chatbot tricked into making heavily discounted, and supposedly legally binding agreements. The episode also highlights AI hallucinations in OpenAI’s Whisper transcription tool, which has been found to fabricate medical transcriptions with potentially dangerous consequences. Gacoscos emphasizes the importance of learning from real-world case studies and plans to feature guest experts to discuss AI security challenges and mitigation strategies.

ChAI Chat Episode 1: Ads Dawson - AI Security and OWASP Top 10 for LLMs

In this episode of the ChAI Chat podcast, host Jomar Gacoscos welcomes Ads Dawson, a Staff AI Security Researcher. They discuss their first meeting at DEFCON, Ads' journey from information security to AI security. Ads shares his insights on his contributions to OWASP and the ethical considerations surrounding AI security, emphasizing the importance of understanding vulnerabilities in AI applications. Ads and Jomar Gacoscos also delve into the complexities of vulnerabilities in LLM applications, particularly focusing on OWASP top 10 vulnerabilities for LLMs, particularly LLM08: Excessive Agency and LLM10: Model Theft. They discuss the implications of Anthropic's new computer use feature and the associated security risks.  Resources and Links Podcast Guest, AI Security Researcher Ads Dawson (aka GangGreenTemperTatum) https://www.linkedin.com/in/adamdawson0/ [https://www.linkedin.com/in/adamdawson0/]  https://github.com/GangGreenTemperTatum [https://github.com/GangGreenTemperTatum]  AI Security Researcher Johann Rehberger (aka Embrace The Red) https://embracethered.com/blog/ [https://embracethered.com/blog/]  https://x.com/wunderwuzzi23 [https://x.com/wunderwuzzi23]  OWASP Top 10 for LLMs https://owasp.org/www-project-top-10-for-large-language-model-applications/ [https://owasp.org/www-project-top-10-for-large-language-model-applications/]  Proof Pudding (CVE-2019-20634) https://avidml.org/database/avid-2023-v009/ [https://avidml.org/database/avid-2023-v009/]  https://github.com/moohax/Proof-Pudding [https://github.com/moohax/Proof-Pudding]  Chapters 0:00 Introduction to the ChAI Chat Podcast 2:47 Meeting at DefCon - A Unique Experience 5:46 Ads Dawson's Journey into AI Security 8:53 Transition into AI Security 11:49 Understand Cybersecurity and AI Security Intersections 14:58 Contribution to OWASP and AI Security Projects 17:36 Exploring Vulnerabilities in AI Applications 23:15 Understanding OWASP Vulnerabilities in LLM Applications 23:53 Exploring Excessive Agency Vulnerability (LLM08) 28:10 Model Theft (LLM10) and Its Implications  34:23 Anthropic's Computer Use Feature and Security Risks 42:54 Community Engagement and Networking in InfoSec