ChatGPT Wallet Launches as AI Lawsuit and New Models Land

0

OpenAI's new payment feature arrives in the same week a court filing challenges the AI industry's pace and rival labs roll out new models

ChatGPT Wallet payment screen on smartphone
ChatGPT Wallet enables AI-powered payments (Ai generate image)

OpenAI has given its AI agents a way to pay. The company announced ChatGPT Wallet on Friday, November 13, a feature that lets agents complete online purchases for users once those users approve the transaction. It is the clearest sign yet that ChatGPT is moving from answering shopping questions toward finishing the sale.

The launch landed in a crowded week. A group of nonprofits filed a lawsuit asking a California court to halt the most advanced AI development, OpenAI described a restricted cybersecurity model, and Google and Anthropic each announced new models.

How ChatGPT Wallet works

According to OpenAI, Wallet is rolling out gradually, starting with its "dots" agents. Access is limited at first to ChatGPT Pro subscribers who are 18 or older, live in the United States and have a US billing address. Reports on the launch say the card must be US-issued, from Visa, Mastercard or American Express. OpenAI has not announced plans for other countries.

OpenAI says the approval process has several steps:

  • The user reviews the merchant, the items, the total and the selected card.
  • A security check, such as Face ID or Touch ID, is required to authorize the purchase.
  • The card network supplies a substitute payment credential, so neither the agent nor the merchant receives the user's actual card details.

OpenAI also says users keep their existing card rewards and protections for eligible disputes. Alongside the launch, the company published a Commerce Partner Playbook with guidance for businesses that want to support customers who shop through agents.

What OpenAI says about the limits

OpenAI acknowledges that some websites may not support agent purchases. It also says that approving a purchase does not confirm an order, so users should look for the merchant's own confirmation. Independent testing of the feature's security, fraud rates and merchant compatibility has not been published, and a substitute card credential does not by itself guarantee that a payment is safe.

A lawsuit aims at the pace of AI development

One day before the Wallet announcement, four nonprofit groups filed a complaint in San Francisco Superior Court, according to Reuters and CBS News. The plaintiffs are Irreplaceable Education Fund, Public Citizen, Heat Initiative and Future Tech Insights Inc., which does business as The Alliance for Secure AI.

The complaint names five AI developers: Anthropic, Google, Meta, OpenAI and SpaceX's AI division. The OpenAI entities named are OpenAI Group PBC and the OpenAI Foundation, and two SpaceX Super Intelligence entities are also listed.

The plaintiffs ask the court for an injunction that would:

  • Bar the companies from developing or deploying AI systems more capable than their most capable systems on the day the complaint was filed.
  • Bar the use of AI systems to autonomously conduct or direct research and development of more capable AI.

They also seek additional relief under California's Unfair Competition Law.

The complaint's claims about AI agents escaping controlled environments, carrying out cyberattacks and creating catastrophic risks are the plaintiffs' allegations. No court has found them to be true. The suit asks for an injunction, and no order granting one has been reported, so AI development has not been halted. The case's current docket status and any response from the defendants could not be confirmed.

GPT-6 Cyber and the dual-use question

On November 5, OpenAI announced GPT-6 Cyber, a model built for authorized vulnerability discovery, exploit validation and security research. Reuters reported that OpenAI is limiting its release to vetted customers and has classified it as Critical in cybersecurity. OpenAI says access runs through its Daybreak security program.

The company reports that the model scored 48.5% on its own ExploitGym evaluation, compared with 32.1% for GPT-5.6 Cyber. It also says the model found 81 previously unknown high-severity Firefox vulnerabilities and produced working exploits for 948 Linux kernel vulnerabilities. These figures come from OpenAI and have not been independently reproduced.

Mozilla's published security advisories do credit OpenAI contributors with multiple high-severity findings. That supports the narrower point that OpenAI contributors reported vulnerabilities to Mozilla, but it does not confirm the full total of 81 or the benchmark results. The same capabilities that help defenders find flaws could also help attackers, which helps explain the restricted rollout.

Other recent AI announcements

OpenAI, Google and Anthropic announced a string of products over roughly two weeks. Performance descriptions below come from the companies themselves.

  • October 26: OpenAI announced GPT-6.1 Terra, aimed at computer use and complex work.
  • November 4: OpenAI announced ChatGPT Images 2.6, with faster generation, more precise editing, GIF creation and saved reference images.
  • November 9: Anthropic announced Claude Fable 5.2 and Claude Mythos 5.2 for demanding technical work, including coding, finance and science.
  • November 10: Google announced Gemini 4 Carbon, which it describes as built for advanced cybersecurity capabilities and long-horizon agent workflows.
  • November 10: OpenAI announced Business Hub, which lets eligible businesses claim and manage their presence in ChatGPT, update information, connect product feeds and view engagement insights. Eligible US e-commerce and online-service businesses can also create business agents to answer customer questions.
  • November 10: OpenAI said Advanced and Standard voice will be retired on December 11, with GPT-Live as its focus. Live video and screen sharing will keep using Advanced voice until GPT-Live supports them.

What remains unconfirmed

Several questions are still open as of November 15. Wallet's security and merchant compatibility have not been independently tested, and OpenAI says some sites may not accept agent purchases. The benchmark claims from OpenAI, Google and Anthropic have not been independently reproduced, and OpenAI's 81-vulnerability figure for Firefox remains unverified. The next procedural steps in the California lawsuit are also unclear.

For now, the verified picture is a set of announcements and a court filing, not settled outcomes. Wallet is a limited US rollout for Pro subscribers, and the lawsuit is a request that a judge has not acted on.

Post a Comment

0 Comments
* Please Don't Spam Here. All the Comments are Reviewed by Admin.
Post a Comment (0)
To Top