Source: The Verge
Part of: Companies put AI agents to work and hit risks →
United's explicit ban on autonomous AI booking reveals friction between enterprise deployment and liability management. The airline is betting that direct human liability and regulatory exposure outweigh the convenience gains of frictionless reservation systems. The move pushes verification responsibility back onto users and their AI tools, fragmenting the seamless agent experience that companies like OpenAI and Anthropic are marketing. It exposes a gap between AI capability demos and real-world enterprise risk tolerance. High-stakes transactions—travel, payment, identity—will remain gated behind human decision-making longer than lower-friction verticals, limiting where agentic automation can operate without explicit airline or brand redesign.