Meta AI AI Vendor Risk Profile

Meta's AI research division developing the Llama family of open-weight foundation models. Largest contributor of open-weight frontier models globally.

Visit Meta AI website

HQ: United States · Frontier Builder

Risk overview

Risk score: 32/100

Risk tier: Moderate

Safety rating: 68/100 (higher is better)

Lower risk scores indicate lower assessed risk.

Last verified: Apr 19, 2026 (107 days ago) Aging · 8 cited sources

Risk dimensions

DimensionRisk score
Data Handling54/100
IP Exposure43/100
Jurisdiction13/100
Security18/100
Regulatory Compliance40/100
Transparency10/100
Business Stability11/100
Dependency ChainNot assessed
Agent GovernanceNot assessed

Analyst summary

Rating: Caution

Meta's Llama family is a leading open-weights option with wide deployment, but Meta AI the consumer product is a data-hungry platform built on top of Facebook and Instagram's surveillance infrastructure. The Kadrey v. Meta training-data lawsuit and the 700M MAU commercial-use cap on Llama weights are material for enterprise use.

Bottom line: Use Llama open weights via a trusted host or self-deployment; avoid Meta AI consumer products for business purposes.

Strengths

Concerns

Best for

Avoid if

Citations

  1. Data Handling — Data Retention Period (primary · medium confidence)
    https://www.facebook.com/privacy/genai
    Verified 2026-04-19
    We retain AI interactions for as long as needed to provide the features and improve our AI models, unless you delete them.
  2. Data Handling — Trains On User Data (primary · high confidence)
    https://www.facebook.com/privacy/policy
    Verified 2026-04-19
    We use information you share on our Products and interactions with generative AI features to develop and improve AI at Meta, subject to regional controls.
  3. Ip Profiles — Commercial Use Permitted (primary · high confidence)
    https://www.llama.com/llama3_3/license/
    Verified 2026-04-19
    If the monthly active users of the products or services made available by or for Licensee exceeds 700 million monthly active users, you must request a license from Meta.
  4. Ip Profiles — Known Ip Lawsuits (official · high confidence)
    https://www.courtlistener.com/docket/66527113/kadrey-v-meta-platforms-inc/
    Verified 2026-04-19
    Kadrey v. Meta Platforms: authors including Richard Kadrey and Sarah Silverman allege Meta used pirated books from LibGen to train Llama models.
  5. Ip Profiles — User Owns Outputs (primary · high confidence)
    https://www.llama.com/llama3_3/license/
    Verified 2026-04-19
    Subject to the terms of this Agreement, Meta grants you a non-exclusive, worldwide, non-transferable and royalty-free limited license to use, reproduce, distribute, copy, create derivative works of, and make modifications to the Llama Materials.
  6. Jurisdiction Profiles — Incorporation Country (official · high confidence)
    https://investor.atmeta.com/financials/sec-filings/
    Verified 2026-04-19
    Meta Platforms, Inc. is a Delaware corporation headquartered in Menlo Park, California (SEC Form 10-K).
  7. Security Compliance — Gdpr Compliant (secondary · medium confidence)
    https://www.facebook.com/privacy/policies/eea
    Verified 2026-04-19
    Meta Platforms Ireland Limited is the data controller for EEA users and has suspended some generative AI features in response to EU Data Protection Authority concerns.
  8. Security Compliance — Soc2 Type2 (primary · medium confidence)
    https://ai.meta.com/trust-center/
    Verified 2026-04-19
    Meta has undergone SOC 2 Type II audits for services covering Meta AI business offerings.