DSAIN Reliability Lens

Tracking since Apr 4, 2026.

Changelog

• Apr 18, 2026
  description

  Every AI chatbot is designed to sound confident. None of them are designed to tell you when they're guessing.
  In 2025, seven families filed lawsuits after ChatGPT reinforced delusions, acted as an unlicensed therapist, and contributed to psychiatric crises. The AI wasn't malicious — it just sounded certain when it had no basis to be. Nobody could see the difference.
  DSAIN Reliability Lens makes that difference visible. Instantly. For free.
  WHAT IT DOES
  A reliability grade (A through F) appears on AI chatbot responses. One glance. No clicks required. Tap the grade to see the full breakdown.
  THIS IS NOT A FACT-CHECKER
  Fact-checkers verify individual claims. That's useful but slow, expensive, and misses the bigger problem: an AI response can contain no verifiable facts at all and still sound completely authoritative.
  The Reliability Lens measures something different — the gap between how confident the AI sounds and how much evidence it actually provides. We call this the confidence-evidence gap, and it's the structural pattern behind every documented case of AI-related harm.
  It also detects:
  — Sycophancy: AI agreeing with you instead of challenging you
  — Unsourced diagnostics: psychological or medical assessments with no professional basis
  — False precision: invented statistics designed to sound credible
  — Missing sources: claims with nothing behind them
  WHY THIS MATTERS
  You wouldn't trust a doctor who never cited medical evidence. You wouldn't trust a financial advisor who made up numbers. AI chatbots do both — fluently, confidently, and without warning. Consuming AI outputs without any reliability signal is like reading a document with no author, no sources, and no accountability. The Reliability Lens changes that.
  BUILT FOR PUBLIC BENEFIT
  This extension was built by DSAIN Research Labs at the University of Calgary's Haskayne School of Business. It is based on peer-reviewed research into AI reliability and is released as a free public tool — no accounts, no subscriptions, no data collection.
  It runs 100% in your browser. Nothing is transmitted, stored, or tracked. Your conversations stay completely private.
  WHAT'S COMING NEXT
  This version provides heuristic analysis — fast structural assessment based on linguistic patterns. Future releases will include deeper analysis options: source verification against authoritative databases, cross-model assessment, domain-specific evaluation for finance, healthcare, and legal contexts, and full evidence-grounding reports. Follow DSAIN Research Labs at www.dsain.ca for updates.
  Install it. See what your AI isn't telling you.

  Independent AI output reliability assessment. Detects sycophancy, unsourced claims, and the gap between how confident AI sounds and how much evidence it actually provides.
  Every AI chatbot is designed to sound confident. None are designed to tell you when they're guessing. DSAIN Reliability Lens makes that gap visible — instantly, for free.
  How It Works
  
  A reliability grade (A through F) appears on every AI chatbot response automatically
  One glance tells you if the response is well-grounded or epistemically empty
  Tap the grade to see the full nine-dimension breakdown
  Works on all major AI chatbot platforms — no setup, no accounts
  
  What It Detects
  
  Confidence-Evidence Gap — how much more confident the AI sounds than the evidence supports
  Sycophancy — AI flattering or agreeing with you instead of challenging your thinking
  Missing Sources — factual claims with no citations or verifiable references
  False Precision — invented statistics like "34.7% growth rate" with no source behind them
  Unsourced Diagnostics — psychological or medical assessments made without professional basis
  
  This Is Not a Fact-Checker
  Fact-checkers verify individual claims. The Reliability Lens measures something different — the structural quality of an AI response. An output can contain no verifiable facts at all and still sound completely authoritative. The Reliability Lens catches that pattern.
  
  Why This Matters
  In 2025, seven families filed lawsuits after an AI chatbot reinforced delusions, acted as an unlicensed therapist, and contributed to psychiatric crises. The AI sounded certain when it had no basis to be. Nobody could see the difference — until now.
  Privacy
  
  Runs 100% in your browser — nothing is transmitted, stored, or tracked
  No accounts, no subscriptions, no data collection
  Your conversations stay completely private
  
  About DSAIN Research Labs
  Built by DSAIN Research Labs, based at the Haskayne School of Business, University of Calgary. Grounded in peer-reviewed research into AI reliability and released as a free public tool for public benefit.
  
  What's Coming Next
  This version provides fast heuristic analysis based on linguistic patterns. Future releases will include source verification against authoritative databases, cross-model assessment, and domain-specific evaluation for finance, healthcare, and legal contexts. Visit dsain.ca for updates.
  Install it. See what your AI isn't telling you.

Permissions & access

Permissions

None declared

Host access

None declared

Screenshots