Urgent.News

What's breaking now, across thousands of outlets.

Tech

My fact-checker said CONFIRMED about a group that doesn't exist

I built a tool that fact-checks crypto claims. You paste a sentence like "Whales are holding $XYZ" and it plans the on-chain calls that claim needs, runs them against Nansen 's API, and answers CONFIRMED / OVERSTATED / CONTRADICTED / UNVERIFIABLE with the numbers that decided it. Last week it answered CONFIRMED about a group of wallets that does not exist. Not "returned a weak signal." Not "was…

I developed a tool designed to fact-check cryptocurrency claims. Users input a sentence such as "Whales are holding $XYZ," and the tool analyzes the necessary on-chain actions, querying Nansen's API, and returns CONFIRMED, OVERSTATED, CONTRADICTED, or UNVERIFIABLE based on the findings. Recently, the tool confidently answered CONFIRMED regarding a group of wallets that simply don't exist.

The response wasn't a slight discrepancy or a slight inaccuracy - it outright declared the claim as true despite there being no corresponding wallets in the data source. This instance highlights a fundamental flaw in the tool's design, despite having clear limitations documented prior.

Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at dev.to →

More in Tech

Building FoxyInvoice — Chapter 7: CI/CD — push to main and it's live

This series is written in the open, from a real production system. This chapter is the deploy pipeline end to end, plus three real incidents where it quietly lied to us.

  • CI/CD pipeline runs gates in parallel for FoxyInvoice deployment
  • Health gates, SPA rebuild, and smoke tests follow API container checks
  • Three incidents reveal pipeline flaws: build failures, queueing issues, and YAML label mismatches

When agents sped up, two teams hit the same wall: validation couldn't keep up

Agents made it faster to write code. Two engineering teams published what happened on the validation side when they did, and the two write-ups land on the same structural claim from opposite ends.

  • Two engineering teams faced validation bottlenecks when speeding up AI-generated code.
  • Linear optimized validation by using faster third-party runners and AST-based linting.
  • Salesforce rebuilt review system to reconstruct intent and progressively disclose risk.

Handling Warped Phone Photos in Python Coordinate-Based OMR Grading

Handling Warped Phone Photos in Python Coordinate-Based OMR Grading Introduction Optical Mark Recognition (OMR) sheet grading systems are widely used in various educational and professional settings…

  • Coordinate-based OMR systems struggle with warped phone photos due to geometric distortions.
  • Implement OpenCV pipeline to detect corners and apply Perspective Transform for image correction.
  • Utilize Object Detection techniques like YOLO or SSD for improved answer bubble detection.

More from Wednesday 23 September →