Fri, 05 Jan 2024 17:10:32 GMT

🚨NEW🚨 After three years investigating public benefits technologies, EPIC has filed an FTC complaint against Thomson Reuters for its development and operation of a faulty benefits fraud system. A 🧵

Across the country, state agencies rely increasingly on private AI vendors to manage public benefits programs. These vendors operate in opaque ways, meaning that agencies sometimes rely on AI decisions they don't understand to determine benefits eligibility.

These vendors aren't tiny, no-name companies. Massive firms like Deloitte, LexisNexis, and Thomson Reuters are all in on the game, making millions off state contracts. I wrote more on this last year: https://t.co/JwkwQ9A9wk

Now, for our complaint: Thomson Reuters (the massive conglomerate behind Reuters, Westlaw, and tons of other services) contracts with several states and D.C. to provide an automated benefits fraud detection system known primarily as "Fraud Detect."

Here's the thing: Fraud Detect doesn't just use data we'd expect for fraud detection. It uses everything from social media data, how far someone travels for groceries, and where someone lives. People are losing public benefits because of their social media profiles!

If you're wondering how that works, it's simple: it doesn't. Reports from California, a state that uses Fraud Detect, suggests that the system is more often than not wrong. Accuracy rates were as low as 46%! (https://perma.cc/98SC-LGYH)

It's time we do something about Thomson Reuters' flagrant abuse of consumer data to peddle faulty AI products that harm those most in need. So EPIC filed an FTC complaint.

Our complaint argues that Thomson Reuters' development and operation of Fraud Detect is unfair under the FTC Act and violates several provisions of the Fair Credit Reporting Act. Thousands of consumers are being harmed, and the FTC has the power to intervene.

This has been 3 years in the making, but it's not the end. The issues we highlight are central to many AI tools today: high error rates, data bias, poor oversight. It's time we stop relying on faulty AI to "fix" our problems and start viewing AI as a problem itself.

Want to read more? Check out EPIC's full complaint here: https://epic.org/epic-urges-ftc-to-investigate-thomson-reuters-fraud-detection-system-in-new-complaint/

My Feed