Sardine

Image Generation just made expense fraud or insurance fraud dead simple, and every company is at risk. This receipt? Generated in 15 seconds with ChatGPT. It has: - Accurate NYC tax rate (8.875%) - Real restaurant and menu data - Authentic-looking receipt format - Embedded EXIF data showing it was "taken" on-location Many classic fraud detection tools are fooled by these images. Now imagine doing this for damaged goods, cars, or even your house. Without physical artefacts it becomes almost impossible to validate an “original” Expect forward-thinking companies to pivot to native photo capture and advanced liveness checks within their apps, which can: 1. Verify and obtain true device location eg from GPS, IP and triangulate 2. identify true location behind a proxy or VPN and establish its the same as where the user is supposed to be 3. Corroborate it with other data points like altitude, air pressure, and see if they match 4. Borrow liveness checks as done in documentary KYC to perceive depth, shadows, etc Is your insurance or expense management system ready for this reality? We offer features 1-3 and many great providers out there offer 4. Hit us up if you are interested in chatting. [A few of you spotted the European style comma's, there's always a hidden tell. Can you spot anything else?]

Return Abuse is shockingly easy to learn and commit. What looks like a “hack” is actually abuse of Amazon’s refund policy. Return abuse keeps coming up as a huge issue whenever we speak to merchants. Fraudulent returns make up 7-10% of all returns, meaning billions in preventable losses, and its a problem thats getting worse. As many e-commerce pro’s know the fraud pattern is simple. - Order one item from Amazon and use it - Order a second from Temu and keep it packaged - Send the second, identical item to Amazon for a refund Now you have the original you bought from Amazon and your refund. But what does the merchant have? They have an impossible task verifying that the item you returned is not the one you received. Some signals we see adding value for our merchant clients - Consistent use of the same device or IP for multiple high-value returns. - Abnormally high transaction amounts compared to purchase history. - Email, phone number or address reputation history (i.e. are they offtending elsewhere) This is why networks are so important. Our Sonar consortium has added returns abuse signals, to our network of over 2.5bn devices, and our clients make up some of the largest merchants, gift card sellers, platforms and marketplaces like StockX, Blackhawk Network, GoDaddy, Square and Checkout. Com.

Announcing: Sardine[Con] - Ending the Scamdemic: Our 1 day event in SF on the 20th August 20205. No more talk. It’s time to take action. We’ve put together a line up of Erin West, Dr Nicola Harding, and of course Soups Ranjan, and Matt Vega, 3CI, SIP, ECFE, CPFPP. We’ll share global best practices on dealing with scams. But then, we need you. Bring your ideas, your ambitions, and your commitment to reducing scams. Let’s do this. Mark the 20th of August as a date for your diaries. The event is on the 20th August 2025 at the Pearl in San Francisco Head to Sardine.ai/sardine-con to register.

Image Generation just made expense fraud or insurance fraud dead simple, and every company is at risk. This receipt? Generated in 15 seconds with ChatGPT. It has: - Accurate NYC tax rate (8.875%) - Real restaurant and menu data - Authentic-looking receipt format - Embedded EXIF data showing it was "taken" on-location Many classic fraud detection tools are fooled by these images. Now imagine doing this for damaged goods, cars, or even your house. Without physical artefacts, it becomes almost impossible to validate an “original” Expect forward-thinking companies to pivot to native photo capture and advanced liveness checks within their apps, which can: 1. Verify and obtain true device location eg from GPS, IP and triangulate 2. Identify true location behind a proxy or VPN and establish it's the same as where the user is supposed to be 3. Corroborate it with other data points like altitude, air pressure, and see if they match 4. Borrow liveness checks as done in documentary KYC to perceive depth, shadows, etc Is your insurance or expense management system ready for this reality? We offer features 1-3 and many great providers out there offer 4. Hit us up if you are interested in chatting.

🤖 AI bots can be used for fraud. If you just took the CDN basic bot detection you're about to be COOKED. Here's why you need better bot detection 👇 Fraudsters are the first to take advantage of new products. Bots have been around for decades. But making them adaptable and sophisticated just got cheaper and faster. Bots can - Steal item descriptions and images to create counterfeit pages - Rapidly create new accounts (new account fraud or NAF). - Cause a spike chargebacks Detecting advanced bots means looking for clues of nonhuman behavior in sessions, or on the client device. Our team has been doing this for decades at giant merchants like Uber, commerce sites like PayPal and in the high seas of digital assets. We can help.

Announcing: Sardine[Con] - Ending the Scamdemic: Our 1 day event at The Pearl in SF on Aug 20, 2025. No more talk. It’s time to take action. We’ve put together a line up of Erin West, Dr Nicola Harding, and of course Soups Ranjan, where we’ll share global best practices on dealing with scams. We’ll share what we have, but then, we need you. The industry, to bring your ideas, your ambitions, and your commitment to reducing scams. Let’s do this. If that’s something you’d be interested in, mark the 20th of August on your calendars. Head to www.sardine.ai/sardine-con to register.

We're excited to share: You can now use the Sardine Issuing API with Model Context Protocol (MCP)! Model Context Protocol (MCP) is a simple way to tell a Large Language Model like Claude how your API or service works. Think of it like the "I know Kung Fu" scene from the Matrix. Here's an example Lets say you're building a Neobank and want to change when and how a card transaction is approved / declined or stepped up for a 2 Factor Authentication In this video example, Kazuki Nishiura shows - A simple API integration - Recommended decision logic for authorizations on payments - API calls ready to copy+paste Build this into your workflow, and build faster. Less searching through docs. Talk to your accounts team for access, or request a demo to see more about how MCP can help you build faster.

Is your insurance or expense management system ready for this reality? We offer features 1-3 and many great providers out there offer 4. Hit us up if you are interested in chatting.

Can't think of the rule to catch that spike in scams? Try building one with AI. When you know what you want to do but you don't know quite how to build it. That's when AI is at its best. For example "Detect if a user shares a device with another customer who has a non-US IP" (pictured below) Say a rule in natural language and Sardine turns it into a platform rule. Want to give it a shot? Come for a demo :)

At Sardine, we’ve always cast a wide net — fraud, compliance, even credit risk. But today, we’re diving deeper. Into... sharks. 🦈 Introducing the Shark Detector 3000™ — powered by machine learning, sonar, and good ol’ fashioned fear. First catch? This ID seemed extremely phishy: Name: Doug Finn DOB: April 1st, 1984 Address: 123 Coral Reef Blvd He claimed to be a dolphin. Our system said otherwise. - Failed Know Your Creature™ (KYC) check - Flagged by FINtelligence AI - Biometrics matched "Great White, Male, 40, Known Scammer" Because even sharks have to follow FinCEN regulations. No fins left unchecked.™