DeepSeek AI Vendor Risk Profile

Chinese AI lab that developed DeepSeek V3 and DeepSeek R1, achieving frontier-level performance at significantly lower training costs. Known for open-weight releases that disrupted pricing expectations across the AI industry.

Visit DeepSeek website

HQ: China · Frontier Builder

Risk overview

Risk score: 67/100

Risk tier: High

Safety rating: 33/100 (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 Handling80/100
IP Exposure55/100
Jurisdiction79/100
Security65/100
Regulatory Compliance80/100
Transparency50/100
Business Stability49/100
Dependency ChainNot assessed
Agent GovernanceNot assessed

Analyst summary

Rating: Avoid

DeepSeek produces capable open-weights models but the hosted consumer app stores data on servers in the People's Republic of China, under PRC jurisdiction including the National Intelligence Law. The Italian Garante has already ordered the app blocked. Self-hosted use of DeepSeek weights is a different risk profile from using the hosted service.

Bottom line: Hosted service is off-limits for most enterprise use; self-hosted open weights is a separate, narrower conversation.

Strengths

Concerns

Best for

Avoid if

Citations

  1. Data Handling — Data Residency Options (primary · high confidence)
    https://chat.deepseek.com/privacy-policy
    Verified 2026-04-19
    We store and process your personal information on servers located in the People's Republic of China.
  2. Data Handling — Third Party Data Sharing (primary · high confidence)
    https://chat.deepseek.com/privacy-policy
    Verified 2026-04-19
    We may be required to share your information with law enforcement, regulators, and government bodies in China pursuant to applicable law.
  3. Data Handling — Trains On User Data (primary · high confidence)
    https://chat.deepseek.com/privacy-policy
    Verified 2026-04-19
    We may use your inputs and outputs to improve our Services, including to train our models, unless you are using the Enterprise API with a no-training agreement.
  4. Governance — Government Ties (secondary · high confidence)
    https://www.nytimes.com/2025/01/27/technology/deepseek-ai-china.html
    Verified 2026-04-19
    DeepSeek operates under China's cybersecurity and national intelligence laws, which can compel cooperation with state authorities.
  5. Jurisdiction Profiles — Incorporation Country (primary · high confidence)
    https://www.deepseek.com/about
    Verified 2026-04-19
    DeepSeek is a Chinese AI company based in Hangzhou, owned by High-Flyer, a Chinese quantitative hedge fund.
  6. Jurisdiction Profiles — Subject To China Jurisdiction (official · high confidence)
    https://www.reuters.com/technology/artificial-intelligence/italy-data-watchdog-blocks-chinese-ai-app-deepseek-2025-01-30/
    Verified 2026-04-19
    Italy's Garante ordered DeepSeek to block its service, citing concerns over data being transferred to and processed in China under PRC jurisdiction.
  7. Security Compliance — Gdpr Compliant (official · high confidence)
    https://www.garanteprivacy.it/web/guest/home/docweb/-/docweb-display/docweb/10097450
    Verified 2026-04-19
    The Italian data protection authority found DeepSeek failed to demonstrate GDPR compliance and ordered the app blocked in Italy.
  8. Security Compliance — Soc2 Type2 (inferred · medium confidence)
    https://www.deepseek.com/security
    Verified 2026-04-19
    No public SOC 2 Type II report is available for DeepSeek as of April 2026.