From Payments Unfiltered | How AI shoppers are forcing merchants to rethink fraud with Jeff Otto · · Primer
“OpenAI and Stripe's new agentic commerce protocol went to market quickly and is missing important fraud signals — in the instant-checkout flow we tested through ChatGPT we were able to put through a synthetic order with a fake email, incorrect billing address and VPN-shifted IP that was approved, while a similar order on the merchant storefront was declined and the account was blocked.”
On , Jeff Otto, CMO at Riskified, spoke about fraud prevention during Payments Unfiltered | How AI shoppers are forcing merchants to rethink fraud with Jeff Otto on Primer.
Jeff Otto, chief marketing officer at Riskified, discussed the implications of agentic commerce—where AI shopping agents act on behalf of consumers—on the Payments Unfiltered podcast in February 2026. He described the trend as "software becoming the customer" and noted that introducing an agent into the transaction flow raises new questions about risk ownership and liability. Otto stated that Riskified tested OpenAI and Stripe's agentic commerce protocol and found it "missing important fraud signals," claiming that a synthetic order with a fake email, incorrect billing address, and VPN-shifted IP was approved through ChatGPT's instant-checkout flow, while a similar order on the merchant storefront was declined. He also commented on Amazon's lawsuit against Perplexity, describing it as "fascinating" and outlining potential future paths for the technology, including an "LLM-centric future" that could create tension for merchants concerned about brand trust. At Riskified's Ascend 2024 user summit in June 2024, Otto announced the launch of Decision Studio and policy decisions, which he said allow merchants to "tailor specific settings and rules around Policy Protect to guide identity resolution and what to do with certain segments." He also expressed pride in his team for organizing the conference and emphasized the value of merchants sharing insights and "trade craft" to combat fraud collectively. Otto noted that Riskified uses adversarial AI to predict potential attacks and that its fraud intelligence team monitors clear web and dark web communities to anticipate emerging threats.