Witness

Rating: 0.3
Author: Entropy

witness.ai

Poor

Top 83%of 296 ai vendor vendors measured

ai vendor⚠ Unverified — automated scan only, no vendor confirmationScanned 9/6/2026Added 9/3/2026

Score breakdown

AI safetypending

Signals planned: model card presence, training data disclosure, published red-team results. Some checks will be partially automated via website scanning; others require vendor submission.

Verification bonus+0/5 bonus pts

Claimed profile

Domain ownership verified by claiming the profile0/5

Security hygiene & transparency8/28 pts

Disclosure policy

Security disclosure page found0/3

Page references vulnerability reporting0/2

security.txt

security.txt file found

Contact field present0/2

Expires field present0/2

Encryption field present0/1

Trust page

Public trust/security page found+5

HTTPS + HSTS

HTTP redirects to HTTPS+3

HSTS header enforced0/2

Source: response headers from witness.ai · Scanned 9/6/2026

Security headers

Content-Security-Policy header0/3

X-Frame-Options header0/2

X-Content-Type-Options: nosniff0/2

Referrer-Policy header0/1

Source: response headers from witness.ai · Scanned 9/6/2026

Governance+0/31 bonus pts

HIPAA

Not found0

SOC 2 Type II

Not found0

ISO 27001

Not found0

Board-level oversight

Requires vendor submission — not yet collected.

ISO 42001

Not found0

SOC 2 Type I

Not found0

Incident historypending

Signals planned: breach-history lookup, incident response quality assessment. This category is coming soon.

Rubric v1.3.0 · 8/28 implemented pts · click a category to see details

Current:Poor

With verification:Fair(+5pts)

Claim this profile to improve your grade →

Score history (5 updates)

↑ improving

v1F → F → F → D- → D-v5

Score history

Grade improved: F → D-

Score +5pts (3 → 8)

trustPagePresence.found added (+pts)

Recomputed from already-known data, not a full rescan

9/6/2026

Scanned 3d ago

AI models used

Self-disclosed · unconfirmedDeepSeek

Found in witness.ai’s llms.txt, captured by Entropy on 9/6/2026. Not vendor-confirmed — the vendor can confirm or correct this from their dashboard.

Show what we found

DeepSeek — “* DeepSeek Security Risks (https://witness.ai/blog-winning-the-game-of-hide-and-deepseek/): Security implications of Chinese-hosted AI models and data sovereignty concerns” (llms.txt)

View the full llms.txt snapshot Entropy captured →