Summary
API
Submitted URL: https://disconfirmai.com/ja/reading/2026-10-20-18
AI Security Verdict
Low Risk
Confidence: 82%
Tech news article about Meta's Muse Wonder benchmarks on the site's own domain; no forms, no impersonation, no threat-intel or malware signals. Only a 48-day-old, unranked domain — a weak prior, not evidence.
Safety Factors (5)
Details
Page Title
Reinforcement Learning Training Binds to Feedback Environments: High Agent Benchmark Scores Don't Equal Unfamiliar Task Ability | DisconfirmAI
Scan Type
public
Domain Name Analysis
Within the commercial generic top-level domain (.com), 'disconfirmai.com' is registered and has no subdomain. Count 12 characters in 'disconfirmai' split between five vowels and 7 consonants. Segmentation suggests 5 words: d, is, confirm, a, i. Average segment length settles at one character. No strong language cues emerged from the frequency lists.
Screenshot

Page Load Overview
Language Analysis
Primary Language
Detection Details
Website Classification
Primary Category
All Detected Categories
Detected Features
Domain & IP Information
| Requests | IP Address | Location | AS Autonomous System |
|---|---|---|---|
| 28 | 47.237.135.51 | Singapore | AS45102Alibaba (US) Technology Co., Ltd. |
| 28 | 1 | - | - |
Detected Technologies2
Content Similarity HashesFor malware variant detection
TLSH (Trend Micro Locality Sensitive Hash)
Specialized for malware detection and similarity analysis
ssdeep (Context Triggered Piecewise Hashing)
Detects similar content even with modifications
sdhash (Similarity Digest Hashing)
High-precisionHigh-precision similarity detection for forensic analysis
These hashes enable detection of similar websites and malware variants by comparing content similarity even when exact matches aren't found.
Image Hashes
Perceptual Hashes
Other Hashes
Scan History
Scan history not available
Unable to load historical scan data