Amazon is opening up the smarts of its Generative AI listing creation to more sellers, announcing today that sellers in France, Germany, Italy, Spain and the UK will now have access to a tool designed to improve their product listings by generating product descriptions, titles and related details.
Additionally, sellers can also “enrich” existing product listings by automatically adding missing information.
The announcement comes nine months after Amazon first revealed plans to offer its generative AI technology to sellers. Amazon hasn't been very forthcoming about its availability by market, but it's likely limited to the U.S. so far; however, the company quietly released the tools in the U.K. earlier this month, according to an Amazon forum post. And in a blog post today, the company said it rolled out the feature to the U.K. and select EU markets “a few weeks ago,” and that more than 30,000 sellers appear to be using at least these AI-enabled listing tools in that time.
Amazon is pitching these new tools as a way to help sellers get their products listed faster.[List Your Products]Go to the page and enter some relevant keywords to describe your product.[Create]With the touch of a button, you can create the basis for a new listing.[Product image]You can also upload photos and create your listing from the tab.
Amazon Marketing Images from Generative AI-Powered Listings Image Credit: Amazon
Amazon will then magically create a product title, bullet points, and description that you can leave as is or edit. However, given the tendency of large language models (LLMs) to hallucinate, it's unwise to post your product listing unchecked. Amazon acknowledges this and encourages sellers to “exhaustively” review their copy to make sure everything is correct.
“Our generative AI tools are constantly learning and evolving,” the company said in a UK forum announcement two weeks ago. “We're actively developing powerful new features to make generated listings even more effective and make it even easier for you to list your products.”
Earlier this year, Amazon also released a new tool that lets sellers create product listings by submitting URLs for their existing websites. It's unclear when Amazon will extend this feature to markets outside of Europe and the US.
Questions about the data
Amazon is well versed in AI and machine learning across its vast e-commerce empire, but bringing any form of AI to the European market raises potential regulatory issues. First, there's GDPR on data privacy, not to mention the Digital Services Act (DSA) on algorithmic risk, and Amazon's online store has been designated a Very Large Online Platform (VLOP) with the aim of ensuring transparency in the application of AI.
Incidentally, Meta was forced last week to suspend plans to train its AI on public posts from European users under pressure from regulators. And Amazon itself has faced the wrath of EU regulators for misusing merchant data, allegedly obtaining non-public data from third-party sellers to benefit its retail rivals. And this month, British retailers filed a £1.1 billion lawsuit against Amazon over similar accusations.
Amazon's recent foray into generative AI is a different proposition, but it would have required training law students on some kind of data. What data, exactly, is unclear. In its original announcement last September, Amazon shared a statement from Robert Tekiela, vice president of selection and catalog systems, that referred to “diverse sources.”
Our new generative AI models enable you to infer, improve and enhance product knowledge at an unprecedented scale, dramatically improving quality, performance and efficiency. Our models learn to infer product information through learned sources, latent knowledge and logical reasoning. For example, they can infer that a table is circular if the diameter is listed in the specifications, or infer the style of a shirt collar from a picture of the shirt.
Robert Tekiela, vice president of selection and catalog systems at Amazon
TechCrunch has reached out to Amazon for comment on these various issues and will update if or when we hear back.