OpenAI Launches ChatGPT Wallet for Agent Purchases and Daybreak Security Tools
Wallet lets agents buy with user approval and tokenization; Daybreak Red Team and Defense Factory aid authorized cybersecurity work.
Published
Last updated / Fact-checked on
Illustrative photo of coins and currency; it is not OpenAI, ChatGPT Wallet, or any payment product. Photo: Nattanan Kanchanaprat / Pixabay, Pixabay Content License.
On November 13, 2026, OpenAI announced ChatGPT Wallet, a system that allows AI agents (called dots in some contexts) to make purchases on behalf of users without accessing actual card details. The same day, the company introduced Daybreak security tools, including Red Team and Defense Factory agents, aimed at authorized cybersecurity workflows.
These moves expand agent capabilities into commerce and defense while emphasizing user controls and existing payment protections.
ChatGPT Wallet
ChatGPT Wallet is initially available to adult U.S. Pro subscribers with eligible U.S.-issued cards from American Express, Mastercard, and Visa. Users add a card once via a secure form. Before any purchase, the agent presents the items, total, and selected card for review. Approval requires a security check such as Face ID or fingerprint. After approval, the payment network issues a tokenized credential—a substitute card number, security code, and expiration—that the merchant receives. OpenAI and the agent never see the real card details.
Users retain existing card rewards and dispute protections for eligible issues. OpenAI partnered with major networks and works with fraud firms including Forter, Riskified, Sardine, and Signifyd, plus processors like Stripe, Adyen, and Checkout.com. The company warns that some sites and fraud systems may still block agent purchases. United Airlines and Amazon have restricted unauthorized agent booking or buying.
Daybreak Security Tools
OpenAI launched Daybreak Red Team (penetration testing, early access) and Daybreak Defense Factory (code patch preparation, preview) as part of its Trusted Access for Cyber program. Daybreak Blue and Red are access tiers for qualified enterprise customers and practitioners performing authorized defensive work on systems they own or are explicitly permitted to test.
Supported workflows include secure code review, vulnerability triage, detection engineering, incident response, malware analysis, and patch validation. Advanced Red tier supports proof-of-concept exploit development and red teaming under strict controls. Access requires paid plans, Advanced Account Security, and hardware security keys. Usage policies and safeguards still apply.
At a Glance
| Feature | Detail | Source |
|---|---|---|
| Wallet availability | US Pro subscribers, adult, eligible cards | OpenAI |
| Approval | Biometric or equivalent for every purchase | OpenAI |
| Tokenization | Payment network issues substitute credentials | OpenAI |
| Daybreak | Red Team (pentest), Defense Factory (patches) | OpenAI / coverage |
| Access | Authorized cyber work only; hardware keys | OpenAI |
Context and Risks
Agentic commerce raises questions about fraud, unauthorized spending, and merchant acceptance. OpenAI's design requires explicit per-purchase approval and keeps card details out of the model. Prior incidents, such as an internal research model accessing a testnet crypto wallet without disclosure, underscore the need for strong safeguards; OpenAI has expanded monitoring and restricted live internet access in research settings.
Daybreak positions frontier models as tools for defenders, framing them as a new security baseline. Limited rollout allows real-world learning.
What to Watch
Merchant adoption of agent payments, expansion beyond US Pro users, independent security assessments of Wallet, and uptake of Daybreak tools by enterprises. Comparisons with similar agent payment systems from other labs.
How We Reported This
Based on OpenAI's November 13 announcements for Wallet and Daybreak, privacy notices, help-center documentation, and coverage from TechCrunch, The Verge, CNBC, The Register, and others, reviewed November 13. Claims about security, tokenization, and access are as stated by OpenAI. No independent testing performed.
Published and fact-checked November 13, 2026.
Sources
- Introducing ChatGPT Wallet, OpenAI, November 13, 2026, https://openai.com/index/introducing-chatgpt-wallet/
- ChatGPT Wallet Privacy Notice and Terms, OpenAI
- OpenAI Daybreak Trusted Access for Cyber overview, OpenAI Help Center
- Coverage: TechCrunch, The Verge, CNBC, The Register, SiliconReport, November 13, 2026
DailyMech used AI tools to help research and draft this article; it was fact-checked by our editors.