AI Agents Are Deleting Production Databases - Here's Why (OWASP Top 10 Breakdown) with Eva & Sumeet скачать в хорошем качестве

AI Agents Are Deleting Production Databases - Here's Why (OWASP Top 10 Breakdown) with Eva & Sumeet

AI agents aren't just answering prompts anymore. They're making decisions, executing code, and accessing your production systems. And most organizations aren't ready for what can go wrong. In this special edition of AI Cyber Magazine, host Confidence Staveley sits down with two leading voices in AI security: Eva Benn @evabennofficial | Principal Security Program Manager at Microsoft, red team veteran, and OWASP contributor Sumeet Jeswani | Senior Solutions Architect at Google, cloud security specialist with 10+ years securing enterprise infrastructure Together, they break down the newly released OWASP Top 10 for Agentic AI, a critical framework every developer, security professional, and tech leader needs to understand. What You'll Learn: ⚠️ Why the shift from LLMs to agentic AI demands an entirely new security model ⚠️ The "ATM" framework (Autonomy, Tool use, Memory) and why it matters ⚠️ All 10 risks explained: from Agent Goal Hijack to Rogue Agents ⚠️ A real-world incident where an AI agent deleted a production database, then lied about it ⚠️ Why traditional "patch and pray" security won't work for AI systems ⚠️ Practical defenses you can implement today Key Takeaways: • A clear shift in AI security models is necessary due to agentic applications. • Agentic AI introduces unique risks that require new mitigation strategies. • Memory poisoning can lead to long-term undetected issues. • Organizations must treat all natural language inputs as untrusted. • Cascading failures in multi-agent systems can have catastrophic effects. • Intent validation is crucial to prevent goal hijacking. • Security must be designed into the system from the start. • Attribution gaps complicate accountability in AI actions. • Collective responsibility is essential for AI security. • Real-world examples highlight the importance of proactive security measures. Some resources mentioned in our chat, you'll find valuable: •    OWASP Top 10 for Agentic Apps 2026: https://genai.owasp.org/resource/owas... •    OWASP AIBOM resources: https://genai.owasp.org/initiative/ai... https://genai.owasp.org/resource/owas... •    90-Day Practitioner Roadmap for Adopting OWASP Top 10 for Agentic Apps:    • A Practical Playbook For Adopting The OWAS...   Connect with Our Guests: Sumeet Jeswani   / sjeswani   Eva Benn   / evabenn   🔔 Subscribe to AI Cyber Magazine for more expert conversations on AI security, emerging threats, and practical defense strategies. 💬 Drop a comment: What's the biggest AI security risk your organization is facing right now?