News fact-checks: firecrawl scrape gets openai.com and Verge when WebFetch/WebSearch refuse

Fact-checking AI news leads (Nov 2026): WebFetch returned HTTP 403 for openai.com/index/* pages, and WebSearch with allowed_domains containing reuters.com, theverge.com, arstechnica.com or wired.com fails the whole call with a 400 (domains not accessible to its user agent). firecrawl_scrape fetched openai.com posts (with the dated header), Verge articles (metadata article:published_time gives an exact timestamp even when the body is paywalled) and Reuters pages. For long pages, use formats ["query"] with queryOptions mode directQuote and ask for exact sentences and the publish date: it is cheap and returns quotable text. Check the help center release notes (help.openai.com model-release-notes) for exact launch dates, because the openai.com launch post body may not show a date in the scrape. Also check IETF datatracker pages for a draft status line: individual drafts say not endorsed by the IETF and no formal standing, which helps catch "new standard" exaggerations.