Packages ride on a conveyor belt during Cyber Monday, one of the company’s busiest days at an Amazon fulfillment center on December 2, 2024 in Orlando, Florida.
Miguel J. Rodriguez Carrillo | Getty Images
Amazon on Wednesday introduced an artificial intelligence agent that will help third-party merchants operate their online businesses.
The company is adding agentic capabilities to Seller Assistant, its AI tool for third-party sellers, meaning the software can take action on a merchant’s behalf with their permission, Amazon said. The update was announced during Amazon’s annual Accelerate conference for sellers in Seattle.
Amazon said tools like Seller Assistant free merchants up to “spend more time focusing on product innovation and customer relationships,” while its generative AI tool handles more tedious operational tasks.
Amazon has released several AI tools for third-party sellers, which account for more than half of all goods sold on the site, such as a product listing generator and an image and video generator for ads.
Dharmesh Mehta, Amazon’s vice president of worldwide selling partner services, told CNBC in an interview this week that 1.3 million third-party sellers have used its generative AI listing tools, which can produce about 70% of what makes up a product listing on its webstore.
“It really gives the seller, in some sense, a team of experts,” Mehta said. “An expert in listing and in pricing and promotions and supply chain, all the things that a small business normally has to either try and learn on their own, hire someone to be an expert, pay someone to be an expert, or sometimes just accept not being that good at, which is not ideal.”
Read more CNBC tech news
The company said its enhanced Seller Assistant goes beyond answering queries and is capable of coordinating inventory orders and business growth plans, as well as implementing fixes for account issues, potentially helping merchants avoid costly suspensions.
Over time, Amazon expects to add more agentic capabilities based on seller feedback, the company added.
Generative AI has evolved from image and text generators to agentic AI tools that can complete multi-step tasks for users with minimal supervision. Outside of its third-party marketplace, Amazon’s AI lab in San Francisco in March released a preview of an agent that can take action in a web browser.
Seller Assistant uses Bedrock, a software tool that lets users access large language models from Amazon and other companies like Anthropic and OpenAI.
The company doesn’t currently plan to charge merchants to use Seller Assistant, Mehta said.
Sellers pay Amazon to access its in-house fulfillment services, account management services and other offerings. It’s become a sizable business for the company, bringing in $40.3 billion in the second quarter.
Amazon last September launched the first iteration of its AI assistant for sellers, codenamed Project Amelia at the time, allowing merchants to troubleshoot issues with their account, get advice on inventory planning and brainstorm listing titles, among other tasks.
OpenAI will not be allowed use the word “cameo” to name any products or features in its Sora app for a month after a federal judge placed a temporary restraining order for the term on the AI startup.
U.S. District Judge Eumi K. Lee granted a temporary restraining order on Monday, blocking OpenAI from using the “cameo” mark or similar words like “Kameo” or “CameoVideo” for any function related to Sora, the company’s AI-generated video app.
“We disagree with the complaint’s assertion that anyone can claim exclusive ownership over the word ‘cameo’, and we look forward to continuing to make our case to the court,” an OpenAI spokesperson told CNBC.
Lee granted the order after OpenAI was sued in October by Cameo, a platform that allows users to purchase personalized videos from celebrities. Cameo filed a trademark lawsuit against the artificial intelligence company following the launch of Sora’s “Cameo” feature, which allowed users to generate characters of themselves or others and insert them into videos.
“We are gratified by the court’s decision, which recognizes the need to protect consumers from the confusion that OpenAI has created by using the Cameo trademark,” Cameo CEO Steven Galanis said in a statement. “While the court’s order is temporary, we hope that OpenAI will agree to stop using our mark permanently to avoid any further harm to the public or Cameo.”
The order is set to expire on Dec. 22, and a hearing for whether the halt should be made permanent is scheduled for Dec. 19.
Sam Altman, chief executive officer of OpenAI Inc., during a media tour of the Stargate AI data center in Abilene, Texas, US, on Tuesday, Sept. 23, 2025.
Kyle Grillot | Bloomberg | Getty Images
OpenAI announced a new tool called “shopping research” on Monday, right as consumers will be ramping up spending ahead of the holiday season.
The startup said the tool is designed for ChatGPT users who are looking for detailed, well-researched shopping guides. The guides include top products, key differences between the products and the latest information from retailers, according to a blog.
Users will be able to tailor their guides based on their budget, what features they care about and who they are shopping for. OpenAI said it will take a couple of minutes to generate answers with shopping research, so users who are looking for simple answers like a price check can still rely on a regular ChatGPT response.
When users submit prompts to ChatGPT that say things like, “Find the quietest cordless stick vacuum for a small apartment,” or “I need a gift for my four year old niece who loves art,” they will see the shopping research tool pop up automatically, OpenAI said. The tool can also be accessed from the menu.
OpenAI has been pushing deeper into e-commerce in recent months. The company introduced a feature called Instant Checkout in September that allows users to make purchases directly from eligible merchants through ChatGPT.
Shopping research users will be able to make purchases with Instant Checkout in the future, OpenAI said on Monday.
OpenAI said its shopping research results are organic and based on publicly available retail websites, and that it will not share users’ chats with retailers. It’s possible that shopping research will make mistakes around product availability and pricing, the company said.
Shopping research is rolling out to OpenAI’s Free, Go, Plus and Pro users who are logged in to ChatGPT.
A Tesla logo outside the company’s Tilburg Factory and Delivery Center.
Karol Serewis | Getty Images
Tesla is trying to get its “FSD Supervised” technology approved for use in the Netherlands. But Dutch regulators are telling Tesla fans to stop pressuring safety authority RDW on the matter, and that their efforts will have “no influence” on the ultimate decision.
The RDW issued a statement on Monday directed at those who have been sending messages to try and get the agency to clear Tesla’s premium partially automated driving system, marketed in the U.S. as the Full Self-Driving (Supervised) option. It’s not yet available for use in the Netherlands or Europe broadly.
“We thank everyone who has already done so and would like to ask everyone not to contact us about this,” the agency said. “It takes up unnecessary time for our customer service. Moreover, this will have no influence on whether or not the planning is met. Road safety is the RDW’s top priority: admission is only possible once the safety of the system has been convincingly demonstrated.”
The regulator said it will make a decision only after Elon Musk’s company shows that the technology meets the country’s stringent vehicle safety standards. The RDW has booked a schedule allowing Tesla to demonstrate its systems, and said it could decide on authorization as early as February.
Last week, Tesla posted on X encouraging its followers to contact RDW to express their wishes to have the systems approved.
The post claimed, “RDW has committed to granting Netherlands National approval in February 2026,” adding a message to “please contact them via link below to express your excitement & thank them for making this happen as soon as possible.” Tesla said other EU countries could then follow suit.
The RDW corrected Tesla on Monday, saying in a statement on its official website, that such approval is not guaranteed and had not been promised.
Tesla didn’t immediately respond to a request for comment.
In the U.S., the National Highway Traffic Safety Administration opened an investigation into Tesla’s FSD-equipped vehicles in October following reports of widespread traffic violations tied to use of the systems.
The cars Tesla sells today, even with FSD Supervised engaged, require a human driver ready to brake or steer at any time.
For years, Musk has promised that Tesla customers would soon be able to turn their existing electric vehicles into robotaxis, capable of generating income for owners while they sleep or go on vacation, with a simple software update.
That hasn’t happened yet, and Tesla has since informed owners that future upgrades will require new hardware as well as software releases.
Tesla is testing a Robotaxi-brand ride-hailing service in Texas and elsewhere, but it includes human safety drivers or supervisors on board who either conduct the drives or manually intervene as needed. Musk has said the company aims to remove human driers in Austin, Texas, by the end of 2025.