Superpower Daily

Superpower Daily

Your reading space

Sign in
Vitality Brings Google-Powered Health AI to U.S. Employers and...European Commission Proposes Barring Chatbots From Simulating R...Spain’s Data Regulator Receives First AI-Agent Breach Notificat...Irregular Finds an AI Coding Agent Replaced Its Own Underlying...McKinsey and SEMI Forecast U.S. Chip Shortfall of Up to 157,000...Huawei Sets 2027 AI Chip Roadmap Alongside New 4,000-Processor...Vitality Brings Google-Powered Health AI to U.S. Employers and...European Commission Proposes Barring Chatbots From Simulating R...Spain’s Data Regulator Receives First AI-Agent Breach Notificat...Irregular Finds an AI Coding Agent Replaced Its Own Underlying...McKinsey and SEMI Forecast U.S. Chip Shortfall of Up to 157,000...Huawei Sets 2027 AI Chip Roadmap Alongside New 4,000-Processor...
Swipe

Research investigation R0916 / policy impact

Sponsored Agents in ChatGPT Ads: Where Does an Ad End and an Action-Capable Agent Begin?

The bounded public record separates ordinary-ad safeguards from Sponsored Agent accountability. It assigns advertisers responsibility for Sponsored Agent content, configurations, Actions, Output, and Ad Materials, but does not publicly specify several agent-level safeguards that ordinary-ad materials describe or that an action-capable experience could require.

Current public editionv1Sep 16, 2026
Read the original analysis
Verified observations
10

10 measured fields

Supported claims
11

11 material findings

Cited sources
8

8 primary or authoritative

Research score
85

Automated topic and evidence score

Interactive figureSponsored Agents in ChatGPT Ads: Where Does an...
CSV JSON
Data status0 verified records across 1 period

Snapshot only. There is not enough history to claim a trend yet.

Verified observationHover or focus any mark for exact valuesLast updated Sep 16, 2026

Version ledger

Frozen public editions

Each edition preserves the records, method, sources, and downloads available at publication time.

  1. v1 / latestSep 16, 202610 records / 8 sources

    Initial public snapshot with 10 records and 8 cited sources.

Coverage note

This synthesis uses the supplied public evidence available through September 16, 2026. Recovering this saved evidence is not a new collection or experiment. The Sponsored Agent program is described as limited beta access, and public documentation may not reflect pilot specifications or onboarding controls.

Dataset ID
spd:sponsored-agents-in-chatgpt-ads-where-does-an-ad-end-and-an-action-capable-agent-begin-ae13eb3a
Stable URL
/research/sponsored-agents-in-chatgpt-ads-where-does-an-ad-end-and-an-action-capable-agent-begin-ae13eb3a
Version
v1
Coverage
2026-09-16
Records
10
Fields
7
Updated

Read the data

The records behind the figure

CSV JSON
Sponsored Agents in ChatGPT Ads: Where Does an Ad End and an Action-Capable Agent Begin? data records
EntityMetricValueUnitObservedSourceTransform
Sponsored AgentAction or output complaint and recovery pathNo dedicated Sponsored Agent Action or Output complaint, reversal, remediation, or recovery procedure is publicly specified.—2026-09-16https://openai.com/policies/ad-tools-terms—
Sponsored AgentAction types with explicit mandatory confirmationNo Sponsored Agent action category has a publicly stated mandatory confirmation rule.action categories2026-09-16https://openai.com/policies/ad-tools-termsCounted clauses in the bounded sources requiring user confirmation for a defined Sponsored Agent action category; none were found.
Sponsored AgentAgent-specific public measurement fieldsNo conversation, Action, confirmation, completion, failure, or recovery metric is specified.specified metric categories2026-09-16https://openai.com/policies/ad-tools-termsCounted explicit Sponsored Agent measurement categories among conversation, Action, confirmation, completion, failure, and recovery; none were found.
Sponsored AgentConversation-level labelingNo explicit persistent label or agent-output separation rule found in the bounded sources.—2026-09-16https://openai.com/policies/ad-tools-terms—
Ordinary ChatGPT adLabeling and separationSponsored label, advertiser identity, placement below responses, and visual separation are explicit.—2026-09-16https://help.openai.com/en/articles/20001047-ads-in-chatgpt—
Ordinary ChatGPT adNamed Ads Manager metricsImpressions, clicks, spend, CTR, average CPC, average CPM, and conversions.metrics2026-09-16https://help.openai.com/en/articles/20001207-ads-in-chatgpt-the-basicsCounted the distinct reporting metrics explicitly listed on the page.
Sponsored Agent advertiserResponsibility allocationResponsible as deemed GPT builder for content, configurations, Actions, Output, and Ad Materials.—2026-09-16https://openai.com/policies/ad-tools-terms—
Ordinary ChatGPT adUser complaint pathThree-dot menu provides Report ad; reporting hides the ad and sends it to OpenAI for review.—2026-09-16https://help.openai.com/en/articles/20001047-ads-in-chatgpt—
Connected Action applicationUser data receivedApplicable portions of user Content and certain information such as country and state may be sent to the Application.—2026-09-16https://openai.com/policies/service-terms—
Ordinary ad advertiserUser information receivedAggregated performance information; direct messages only when the user chooses to message the advertiser.—2026-09-16https://help.openai.com/en/articles/20001047-ads-in-chatgpt—

Measurement technique

How to read this report

  1. 01Planned evidence matrix: code each control by format—ordinary ChatGPT ad and Sponsored Agent—as explicit, inherited, discretionary, prohibited, or not publicly specified.
  2. 02Use the bounded OpenAI terms, help pages, policies, app terms, and Sponsored Agents program form supplied in the evidence packet.
  3. 03Separate documented rules from absences in the bounded public record; do not treat an undocumented safeguard as proof that it is absent in implementation or private pilot materials.
  4. 04Count only explicitly named ordinary-ad reporting metrics and explicitly mandatory Sponsored Agent Action-confirmation categories.

Sources

Evidence

2 publishers supporting 10 records. Expand a publisher to inspect its cited pages.

openai.comopenai.com6 sources / 6 records
App Developer Terms0 supporting recordsopenai.com0 supporting recordsopenai.com5 supporting recordsopenai.com1 supporting recordopenai.com0 supporting recordsSponsored Agents with HubSpot0 supporting records
help.openai.comhelp.openai.com2 sources / 4 records
help.openai.com3 supporting recordshelp.openai.com1 supporting record
Next report / 01AI Model Economics Index All research reports