To deal with the increase in fake reviews and scams on e-commerce platforms, Mozilla is launching Fakespot Chat, an AI in the form of a Firefox extension that analyzes product pages to give you the best advice.
On the Internet, online shopping can quickly become a real headache. Is the product solid? Does it keep its promises? Are the advertised features accurate? Is it consistent with the photograph? What are its real qualities and weaknesses? So many questions we face when going to checkout. And, unfortunately, this is not improving, with fake reviews abounding – especially since generative AI like ChatGPT now makes it possible to write fraudulent comments on a mass basis – and scams multiplying. Scammers use fake reviews to push their products up in search results and gain customer trust to trigger a sale.
The Mozilla Foundation, which publishes the Firefox browser, is working to implement solutions. Last October, we learned that the company was testing Review Checker, an AI-powered function indicating the degree of veracity of comments on a product page on Amazon, BestBuy and Walmart (see our article). This time, Mozilla announces in a statement its own artificial intelligence, called Fakespot Chat, which is able to see if the reviews on the item are false or biased, or if the device is a counterfeit, all with the aim of reducing search time. A sort of store salesman in short, from whom consumers would ask for advice and who would guide them in their choices.
Fakespot Chat: an AI against fake reviews
To develop Fakespot Chat, Mozilla collaborated with the start-up Fakespot, which the company acquired earlier this year. Specializing in the fight against disinformation and fraud, Fakespot “uses sophisticated artificial intelligence (AI) and machine learning (ML) to detect patterns and similarities between reviews to flag those that are most likely to be misleading”Mozilla explained when announcing the acquisitionannouncing from the outset that it planned to integrate the technology into its browser to “make Firefox clients [internautes les] better equipped to cut through misleading reviews”.
Initially, Fakespot was available as a mobile application on Android and iOS, a web page as well as an extension. Until now, you had to submit an Amazon, eBay or other product to the tool, and it analyzed it by providing various information, including a grade from A to F providing information on the reliability of the product. Since then, Mozilla has developed its own language model (LLM), and Fakespot is the first tool to take advantage of it. The founder and director of Fakespot, Saoud Khalifah, explains that “the goal with Fakespot Chat is to reduce your product research time and help you make better purchasing decisions”.
To take advantage of the chatbot, you must first analyze the product through the Fakespot extension or by going to the web page. Once the analysis is completed, Fakespot Chat appears on the right side of an analysis page with main information such as Fakespot review scores, pros and cons of the product in question. Then just ask him questions, and that’s it.
Fakespot Chat: a standoff with Amazon
Fakespot’s technology uses artificial intelligence and machine learning (machine learning) to sort real and fake reviews. Mozilla says that, mindful of potential security risks, the extension does not store any personal data. “It reinforces what we believe shopping should look like, in a world where a paradigm shift is happening in consumer technology: privacy, security and openness are imperative to our shopping experiences. as individuals on the Internet“, explains the organization. It is the users themselves who refine the chatbot, thanks to “thumbs up” and “thumbs down” buttons. For the moment, Fakespot Chat is only available to buyers from Amazon.com in the United States.
It remains to be seen how the e-commerce giant will react. If he is also in open war against fake reviews and scams on his site, Fakespot had already annoyed him quite a bit in the past. In 2021, Jeff Bezos’ firm declared that the tool was wrong 80% of the time on false evaluations and that it provided “Misleading information about our sellers to customers and creates potential security risks”. He then obtained the removal of the application from the App Store and bought the search results on the keyword “fakespot” in the Apple application store in order to make an Amazon ad appear first. Not sure he appreciates the new initiative…