November 2025 (2025-11) monthly accomplishments for vercel/ai. Key feature delivered was the OpenAI web_search external_web_access control, introducing an external_web_access parameter to allow users to choose between fetching live content or using cached results. This strengthens security governance and improves reliability by enabling explicit guardrails on how data is fetched for web searches. Major bugs fixed: none reported or tracked in this month’s scope for vercel/ai. Overall impact: provides data-access guardrails, reduces potential data leakage, and enables reproducible results for downstream users. Sets the foundation for safer live data access in AI tooling and downstream workflows. Technologies/skills demonstrated: API surface design and parameterization, secure-by-default considerations, OpenAI web_search integration, version-controlled feature delivery, collaboration evidenced by changes in vercel/ai with a clean commit implementing the new parameter. Business value: enhanced control over data freshness and source of truth, reduced security risk for live data fetches, improved consistency and predictability of AI-driven search results, and smoother onboarding for customers requiring stricter data governance.