A June 2026 Riskified report titled "Rewriting the Rules on Returns," conducted by eTail Insights across 2,091 consumers in seven countries, finds that 50% of shoppers use AI tools such as ChatGPT and Claude to draft return or refund claims. Retailers surveyed say these AI-drafted messages produce "highly convincing return requests" that overwhelm review teams. The report also finds that 85% of consumers accept at least one type of strategic or deceptive return behavior as normal, 46% consider it acceptable to return items that look different in person, 42% practice "bracketing" (buying multiple sizes to return unwanted ones), and 24% say it is acceptable to wear an item before returning it. Notably, 52% of consumers support stricter return policies and more than 60% say they would change their behavior if they understood the true cost of returns. "When half of consumers are using AI to draft highly persuasive refund claims, manual reviews simply cannot keep up," said Jeff Otto, CMO at Riskified.
What happened
A June 2026 report titled "Rewriting the Rules on Returns," commissioned by Riskified (NYSE: RSKD) and conducted by eTail Insights, surveyed 2,091 consumers across seven countries (51% US, 14% UK, 12% Japan) and interviewed senior leaders at major retail and e-commerce brands. The central finding: 50% of consumers report using AI tools such as ChatGPT or Claude to draft return or refund claims. Retail leaders interviewed say AI-drafted messages produce "highly convincing return requests" that are overwhelming their review teams. Riskified notes that between 32% and 49% of consumers across surveyed markets already say they prefer AI for initiating a return or exchange, suggesting the 50% adoption rate may continue to climb.
Consumer behavior breakdown
85% of consumers accept at least one type of strategic or borderline return practice as normal - only 15% say they prefer to adhere strictly to merchant policies. Specific behaviors: 46% say it is acceptable to return an item because it looks different in person than it did online; 42% practice "bracketing" - buying multiple sizes or colors to try at home, intending to return most of them; and 24% say it is acceptable to wear or use an item before returning it. Social media is a normalizing force: over half of consumers report encountering return-related "life hacks" on social platforms. For context, a separate 2024 Riskified survey found 51% of consumers admitted to bracketing specifically - a narrower metric than the broader 85% who now accept at least one form of strategic return behavior in 2026.
Merchant response
Retailers are tightening policies in direct response. Practices reported include narrowing acceptable return reason categories (in some cases from over ten categories to five), shortening return windows, requiring returned items be completely unopened and unused, and beginning to deploy AI-powered behavioral detection to distinguish loyal customers from policy abusers. Riskified frames the shift toward identity-based, tiered return policies as the emerging standard - an approach 56% of consumers say they prefer over uniform policies. Zappos is cited as an example of a merchant maintaining liberal free-return policies despite the broader trend.
Consumer attitudes
Despite normalizing certain strategic behaviors, consumer sentiment is not uniformly anti-merchant. 52% of consumers support stricter return policies, and more than 60% say they would adjust their behavior if they better understood the true cost of returns. "The democratization of generative AI has fundamentally changed the returns landscape. When half of consumers are using AI to draft highly persuasive refund claims, manual reviews simply cannot keep up," said Jeff Otto, Chief Marketing Officer at Riskified. "To protect margins without alienating top shoppers, retailers need the ability to provide differentiated return experiences, accurately and at scale."
What to watch
For practitioners building fraud-detection systems: monitor Riskified and competitor platform updates on identity-based return scoring and behavioral signals specific to AI-generated claim text. Track whether the 50% AI adoption rate for return claims continues rising in follow-up surveys, and watch for merchant disclosure of new AI-detection tooling in their returns workflow.
Scoring Rationale #
Riskified's report documents 50% AI adoption for return claims among 2,091 surveyed consumers - a quantified, multi-country finding relevant to fraud-detection and e-commerce operations practitioners. The story is vendor-commissioned (NYSE: RSKD) but based on a real consumer survey corroborated by independent trade coverage; impact is solid for operational AI tooling and policy design, below frontier-research or major model-release tier.
Practice with real Retail & eCommerce data
90 SQL & Python problems · 15 industry datasets
250 free problems · No credit card
See all Retail & eCommerce problems