The Risk of AI adoption and business implication of not using AI
summary
SplxAI offers automated adversarial testing and compliance automation for production LLMs, claiming to have found a multimodal chatbot breach that caught 5,000 customer records and Azure keys plus sensitive files in a live system. Clients include Infobip and major banks; use cases are customer- and employee-facing chatbots, healthcare avatars and voice agents. Core problems are prompt injection, hallucination and bias (IKEA case), plus multimodal/voice attack surface. Recommended approach is shift-left red teaming before deployment, then continuous runtime monitoring and prioritized remediation.
AI-written summary · about
chapters (15)
- 0:00AI adoption risks
- 1:08Speaker background and company origin
- 2:11AI efficiency in healthcare retail
- 3:34AI use cases in Europe
- 4:37Regulatory compliance challenges
- 6:22AI security and safety risks
- 8:28AI security threats
- 9:30Agentic AI attack surface
- 10:36Multimodal AI risks
- 12:03AI market trends competition
- 13:50AI implementation roadmap
- 15:14AI and security teams
- 16:38Shift left in AI security
- 17:41AI adoption and security summary
- 18:24Q&A
related talks
57mAI Applications Security Puzzle
14mAI Sprawl: The Hidden Risk Threatening Your Enterprise Now
37mFundamentals of AI safety and security
52mThe Risky Business, of AI Illiteracy
54mA Practical Guide to AI Threats & Mitigation Strategies
22mFireside Chat - AI Security Vulnerabilities
45mDeploy First, Secure Never: AI Between Productivity and Security Chaos
24mHow to Build Trustworthy AI