United Airlines禁止AI代理访问其订票系统, citing cybersecurity risks。
United Airlines has implemented a blanket ban on AI agents—including RPA bots, marketing automation tools, and custom-built assistants—from accessing its flight booking and itinerary management systems, citing cybersecurity risks as the sole justification. This is the first airline-wide, publicly stated policy of its kind, applying equally to enterprise SaaS, brand-owned agents, and individual developer tools. For early AI application teams, this is a critical cost signal: when core business systems actively close off automation access, the traditional 'integrate-wrap-distribute' model collapses. Founders must shift strategy—away from fragile direct API integrations and toward resilient, compliance-aware architectures. Example: deploy lightweight agents focused solely on intent understanding and structured request generation, then route validated actions through secure, human-in-the-loop channels (e.g., ticketing APIs with manual approval gates or templated email engines). Yes, latency increases—but it avoids the far higher hidden costs of sudden API deprecation, escalating anti-bot measures, and regulatory exposure.