Episode Details

A records-led investigation into the rise of Indirect Prompt Injection (IPI) and the automation of AI exploit engines. Following a thirty-two percent surge in malicious web-based injections reported by Google, Margaret Ellis traces the archival history of how AI agents lost the ability to distinguish between data they read and instructions they must follow.

Topics Covered

• 📋 The thirty-two percent increase in malicious IPI activity between November twenty-twenty-five and February twenty-twenty-six.
• 🔬 Technical analysis of 'invisible injections' including the Tweety bird and flying squid case studies.
• ⚖️ The shift from human-paced attacks to autonomous AI exploit engines.
• 🔍 The relocation of liability as AI agents accumulate authority without corresponding oversight.
• 🛡️ The limitations of OWASP security frameworks in the face of machine-paced reasoning.

Neural Newscast is AI-assisted, human reviewed. View our AI Transparency Policy at NeuralNewscast.com.