<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>Seerror Research Papers</title>
    <link>https://www.seerror.com/research.html</link>
    <description>Open-access technical reports from Seerror Technologies, Bengaluru. On-device AI, scam prevention, and safety evaluation.</description>
    <language>en-in</language>
    <atom:link href="https://www.seerror.com/public/papers/feed.xml" rel="self" type="application/rss+xml"/>
    <item>
      <title>Seerie: An Offline 1.7B-Parameter Assistant for Scam Prevention in India, and a Manual Audit of Its Failures</title>
      <link>https://www.seerror.com/research/seerie-offline-scam-prevention.html</link>
      <guid isPermaLink="true">https://www.seerror.com/research/seerie-offline-scam-prevention.html</guid>
      <pubDate>Fri, 02 Oct 2026 00:00:00 +0000</pubDate>
      <dc:creator>Jay Tiwari</dc:creator>
      <author>seerrorworks@gmail.com (Jay Tiwari)</author>
      <description>Offline 1.7B scam-prevention assistant for India. Published on Zenodo (10.5281/zenodo.23110076). Manual audit 82.9% safe on 70 prompts. Automatic harness 90%, Cohen's kappa 0.10.</description>
      <enclosure url="https://www.seerror.com/public/papers/seerie-offline-1.7b-scam-prevention-india.pdf" type="application/pdf" length="602690"/>
    </item>
  </channel>
</rss>
