Online product reviews are becoming a battlefield for modern AI

The landscape of online <a href="https://shoppixy.com/welcome-to-the-filter-us-the-guardians-home-for-product-reviews-and-recommendations/" title="Welcome to the Filter US, the Guardian’s home for <a href="https://shoppixy.com/how-to-spot-ai-fake-product-reviews/" title="How to Spot AI Fake Product Reviews“>product reviews and recommendations”>product reviews has become a digital battleground where artificial intelligence is pitted against itself.

Generative AI capable of producing human-like reviews is now being challenged by sophisticated software designed to identify fraudulent feedback. This technological clash carries significant consequences for both shoppers and the future integrity of online content.

Saoud Khalifah, founder and CEO of Fakespot, a startup that utilizes AI to spot deceptive reviews, noted that his firm has observed a surge in AI-generated fake content. Fakespot is currently developing methods to identify this material.

“The thing that is very different today is that the models are knowledgeable to a point where they can write about anything,” he said.

While fake reviews have existed nearly as long as the concept of online feedback, the issue has gained new urgency due to the widespread availability of advanced AI tools.

Following years of managing the problem through individual enforcement actions, the Federal Trade Commission proposed a new rule last month aimed at curbing fraudulent reviews. If enacted, the regulation would prohibit the creation of fake reviews, payment for reviews, the suppression of honest feedback, and other deceptive tactics, while imposing substantial fines on violators.

However, defining exactly what constitutes a fake review has become increasingly difficult, and the technology intended to catch such fraud remains a work in progress.

“We don’t know — really have no way to know — the extent to which bad actors are actually using any of these tools, and how much may be bot-generated versus human-generated,” Michael Atleson, an attorney in the FTC’s Division of Advertising Practices, said. “It’s really more of a serious concern, and it’s just a microcosm of the concerns that these chatbots are going to be used to create all kinds of fake content online.”

Evidence suggests that AI-authored reviews are already prevalent. CNBC reported in April that certain Amazon reviews contained obvious signs of AI involvement, with many beginning with the phrase, “As an AI language model …”

Amazon is among the many retailers that have spent years fighting fraudulent reviews. A spokesperson stated that the company processes millions of reviews weekly and proactively blocked 200 million suspected fake entries in 2022. The firm employs a mix of human investigators and machine learning models that evaluate data points such as user review history, login patterns, and connections to other accounts.

The situation is further complicated by the fact that AI-generated reviews are not strictly prohibited by Amazon’s policies. An Amazon representative noted that the company permits AI-assisted reviews provided they are authentic and adhere to all other guidelines.

The e-commerce giant has also signaled a need for external support. In June, Dharmesh Mehta, Amazon’s vice president of worldwide selling partner services, wrote in a company blog post that there is a need for greater cooperation between “the private sector, consumer groups, and governments” to tackle the rising tide of fake reviews.

The central question remains whether detection software can successfully outmaneuver the AI used to generate fake content. Khalifah mentioned that the first AI-generated reviews identified by Fakespot originated from India several months ago, created by what he describes as “fake review farms”—commercial operations that mass-produce fraudulent feedback. Generative AI makes this process significantly more efficient.

“It’s definitely a hard test to pass for these detection tools,” said Bhuwan Dhingra, an assistant professor of computer science at Duke University. “Because if the models are exactly matching the way humans write something, then you really can’t distinguish between the two. I wouldn’t expect to see any detector passing the test with flying colors any time soon.”

Multiple studies indicate that humans struggle to identify reviews written by AI. Numerous companies and researchers are developing systems to flag AI-generated text, with organizations like OpenAI, the creator of ChatGPT, even working on tools to detect their own AI outputs.

Ben Zhao, a professor of computer science at the University of Chicago, argued that it is “almost impossible” for AI to effectively eliminate AI-generated reviews, as bot-written content is often indistinguishable from human writing.

“It’s an ongoing cat-and-mouse chase, but there is nothing fundamental at the end of the day that distinguishes an AI-created piece of content,” he said. “You’ll find systems that claim that they can distinguish between texts written by humans versus ChatGPT text. But the techniques underlying them are all fairly simple compared to the thing that they’re trying to catch up to.”

With 90% of consumers reporting that they consult reviews before making online purchases, the trend has alarmed consumer advocates.

“It’s terrifying for consumers,” said Teresa Murray, who directs the consumer watchdog office for the U.S. Public Interest Research Group. “Already, AI is helping dishonest businesses spit out real-sounding reviews with a conversational tone by the thousands in a matter of seconds.”