Seekr’s Rob Clark on building trust in autonomous systems

Seekr President Rob Clark discussed the security implications of open-weight AI models in defense. He likened these models to software supply chain risks, highlighting vulnerabilities like data poisoning and backdoors. As organizations shift toward smaller, more compact models, verifying provenance and data lineage becomes critical. Clark also emphasized the necessity of an “AI checkride” to validate autonomous systems before deployment in disconnected battlefield environments. To ensure reliability, defense organizations must use AI-driven tools to verify model behavior and train systems to detect sophisticated spoofing or manipulation attempts during high-consequence missions.