Back to AI Tools
Grok
Risk Overview
Moderate Risk
5
/10
AI Agents
·
Conversational AI
·
Image Generation
·
Research
·
Top Risks
Mandatory arbitration with class-action and jury-trial waivers and liability capped at $100 USD creates unbalanced legal exposure for consumers. Active federal litigation (Skaggs v. X.AI, pending as of June 2026) alleges real-time sharing of chat content to Google, Meta, and TikTok without user consent, beyond disclosed service-provider sharing.
Mitigations
Users can disable 'Improve the Model' setting to opt out of model training for future conversations (non-binding control; prior training is not retroactively undone). Grok maintains SOC 2 Type II certification and an active HackerOne bug-bounty program, reducing security-posture risk; no breaches have been publicly disclosed.
Get started
Is your team using
Grok
safely?
Highest Risk Categories
The highest-scoring risk categories identified in this assessment.
Legal Terms
Mandatory arbitration or exclusive Texas state/federal court jurisdiction (Wichita or Tarrant County). Class action and jury trial waivers explicitly stated. Liability capped at amount paid or $100 USD, whichever is greater. One-year limitations period for federal claims, two years for state claims. Arbitration provisions extend to corporate affiliates (SpaceX, X, Cursor entities). EU consumers subject to Irish law and courts with different terms (Release and Indemnity sections do not apply to EU consumers), but consumer tier ToS does not distinguish by region by default.
8
/10
Training Use
Opt-out default: User Content is used for model training unless user disables 'Improve the Model' setting in logged-in account; ToS grants irrevocable, perpetual, worldwide license for any purpose including product development and model training. Opt-out applies to future conversations only; prior training is not undone. Comparable to consumer Claude (score 2, but 2025 shift to opt-out) and riskier than Amazon Bedrock (score 1, no training default).
5
/10
3rd-Party Sharing
Privacy Policy discloses sharing with contracted service providers. Active federal class-action lawsuit (Skaggs v. X.AI, 5:2026cv04550, filed June 2026, pending) alleges real-time sharing of user chat content to Google, Meta, and TikTok via embedded tracking technology. Illinois BIPA class actions filed October 2026 allege biometric data manipulation/sharing. Policy contains broad business-transfer carve-out; no binding opt-out mechanism for third-party analytics/ad-partner access to chat content. Comparable to AskYourPDF (score 5).
5
/10
Unlock the complete assessment
Run a free assessment to access the full AI tool risk analysis.
Updated July 2026 • AI-assisted research • Reviewed by Verax • Report an issue
Reference Documentation
Official documentation and policies referenced as part of this assessment.
Terms of Service
https://x.ai/legal/terms-of-service
Privacy Policy
https://x.ai/legal/privacy-policy