ChatGPT gives publishers’ AI sales pitch a new test
Research and verification notes
11 November 2026. Author: Ivan Massow.

The news peg is OpenAI’s official 10 November 2026 beta launch. Reports dated 6 and 9 November described an unreleased tool; they are superseded on launch status by the official announcement and current setup documentation. The Daily Mail’s test was reported on 26 October, not a new November launch. Future’s client and renewal information was reported on 1 October. Campaign claims are attributed, not independently audited outcomes.

The 50% to nearly75% result describes visibility across the Daily Mail’s selected prompts for an unnamed finance brand. It is not a probability forecast, population reach or measured sales gain. The 100-answer example is explicitly hypothetical. Recommendations about testing, costs and editorial trust are editorial analysis, not claimed company practices or measured failures.

OpenAI’s claiming offer is distinct from Business Agents, which initially cover eligible US ecommerce and online service businesses. Help documentation qualifies availability by location, type and verification support. Platform mention definitions, privacy thresholds and missing data limit comparisons with external prompt panels. No product registration or account change was performed for this report.

The terms discussion is confined to section6 of Businesses in ChatGPT Terms, updated9November, and its permitted-purpose/service-provider exception. It does not claim all client reporting is forbidden or that OpenAI has prohibited consultancy. No publisher’s infringement or breach is alleged.

Noah research limitation:
A normal deep trend-report commission was attempted, with its existing automatic forecast workflow unchanged. Admission stopped before any child retrieval or forecast began:
status: programme_admission_failed
reason: programme_manifest_component_request_invalid
failing_component_id and shape: editorial_discovery
field_path: time_window
validation_error: invalid_params
retryable: false; automatic_retry_safe: false; do_not_poll_again: true.
Plan reference: nabdl_plan_c8b7ed4e6545188f9b92b98cc10ecd7e82c5db654dbad700
Programme reference: nabdl_programme_c8b7ed4e6545188f9b92b98cc10ecd7e82c5db654dbad700
No durable job IDs were created. The returned recovery called for correction in the authoritative server plan; no server-plan editing capability was available. No components were dispatched separately, no forecast questions or replacements were commissioned, and no probabilities were invented. The article uses independently checked original reporting and primary documents. It is not described as completed Noah research. The full admission diagnostic is privately archived.

Quotations:
Farhad Divecha, Accuracast group chief executive: “If their estimation data was wildly erroneous, that will make us rethink paying for those tools”. Krystal Scanlon, Digiday,10November2026. Conditional response to platform data, not proof supplier estimates failed.
https://digiday.com/marketing/openais-new-business-agents-give-brands-a-free-way-into-the-conversation-its-also-selling-as-an-ad/

Mike Peralta, Future chief revenue officer: “You want to make it simple for the brand to understand and buy”. Quoted by Sara Guaglione in Digiday,1October2026, describing his Digiday Publishing Summit remarks. A commercial aspiration, not evidence of campaign outcomes.
https://digiday.com/media/media-briefing-publishers-are-turning-geo-from-an-experiment-into-a-business/

Neither quotation is from a Tomorrow’s Publisher interview. Complete source records, checks, commission, admission diagnostic, cartoon prompt and release receipt are kept privately. The cartoon is a metaphor for incomplete measurement, not a claim that any named campaign had zero sales.