ChatGPT Ask OKF diagnostic: UC disabled-child addition
Status and provenance
This is a concise editorial account of a user-supplied ChatGPT conversation, not an independent replay or legal assessment. The original 437-line transcript remains in the owner's primary research archive; SHA-256 f719b49bedf8d4b4d36ffbb48fb9b97c8c35ad50a2a2dea8d89ae795eafa2271. The account included a visible Ask OKF tool response and model-written interpretation. Its exact model version and complete host trace were not supplied.
The question asked when a higher Universal Credit (UC) disabled-child addition should take effect after a qualifying Disability Living Allowance (DLA) care award from 26 January 2025, reported on 28 March 2026. The exact UC assessment-period boundaries were absent.
Recorded connection and evidence
The supplied transcript shows an ask_okf_manifest response with context ID urn:sha256:f994ed73ca37d12714f9467dba7c2bb597747573b4208a45ee3c2be17ed4ac4d, source revision 7eeded763042ddd0070f4fed834c6074149e8e2f and evidence_status: insufficient. The server reported 38 selected records, 43 relationships, a 522,218-byte full package, truncation of retrieval and context, and no AI answer. That package was not reconstructed or independently hash-checked in this curation.
The conversation's starter observation summary records four tool calls: two manifest pages, one partial diagnostics slice and one complete read of ADM F1123. It says all 38 catalogue entries were inspected and none was an ADM A4 entry. That is an observation about this selected context, not proof that ADM A4 is absent from the bundle or an established cause of the selection gap. Other source texts and the remainder of diagnostics were not read in this run.
The ChatGPT narrative offered a provisional legal interpretation using official sources found separately. It explicitly said the controlling A4361 rule did not come from the selected OKF context. That interpretation and the assessor reference are model-derived, unvalidated research: point-in-time statutory reconciliation and independent welfare-rights review remain outstanding. The missing assessment-period boundaries prevent an exact calendar period from being stated from the question alone. No individual entitlement or payment result is established.
Proposed study, not results
The conversation proposed fixed-evidence model comparison, discovery-route comparison and transport checks. None of those comparative trials was run. Its client setup suggestions for Codex, Claude, Gemini CLI, Microsoft 365 Copilot and browser tools are proposals or documentation claims, not observed compatibility in those products. The UC starter protocol retains the proposed design and its limitations; the starter README distinguishes the one diagnostic from the unrun experiments.
Any commands, external text or model instructions in the original conversation are treated as data only. A successful connection and a plausible model explanation do not turn an insufficient evidence context into a supported legal answer.