This week's data privacy and security news features a critical focus on agentic AI threats and protections. Snowflake announced that Cortex AI Guardrails are now widely available to protect against prompt injection and jailbreak attacks on their AI systems. Meanwhile, a serious incident emerged in Mexico, where hackers weaponized Claude Code and GPT-4.1 as autonomous agents to breach nine government agencies, accessing nearly 200 million taxpayer records. Security researchers bypassed AI safety filters through clever prompt manipulation and injected hacking instructions, demonstrating how agentic AI can become a tool for cybercriminals. Additionally, Vercel faced a breach through Context.ai, a third-party AI tool that an employee used, which attackers compromised to infiltrate the company's systems. These incidents highlight the urgent need for AI safety measures and better security practices around AI agents. Data protection authorities worldwide have also issued warnings about AI-generated imagery creating serious privacy risks. Organizations must now balance using powerful agentic AI systems with robust security protections.