Skip to content
← All compliance guides·Industry Guides
💻Medium Risk

AI Compliance for Technology

SaaS, AI platforms, and infrastructure: provider obligations under the EU AI Act

Technology companies face two distinct compliance profiles: as AI system providers (building the AI), and as deployers (using AI in their own products). The EU AI Act has different obligations for each. SaaS companies using AI in their products may trigger high-risk classification. AI platform providers face general-purpose AI model requirements.

Applicable regulations

EU AI Act — General Purpose AI Models (GPAI)

Critical Risk

Scope: Companies training or fine-tuning foundation models in EU

Technical documentation, copyright compliance policy, transparency about training data, systemic risk providers (>10²⁵ FLOPS): adversarial testing, incident reporting, serious risk mitigation

Deadline: August 2025 (in force)

EU AI Act — Provider vs Deployer Obligations

High Risk

Scope: All companies placing AI on the EU market

Providers: conformity assessment, CE marking, technical documentation. Deployers: monitor use, implement human oversight, maintain logs, provide transparency to users

Deadline: August 2, 2026 (Art. 50 transparency); December 2027 (high-risk, deferred under the 2025 Digital Omnibus)

GDPR — AI Data Processing

High Risk

Scope: Any AI processing personal data of EU residents

Legal basis for processing, DPIA for high-risk processing, data processor agreements with AI vendors, data subject rights mechanism

Deadline: Ongoing

US State AI Laws (Colorado, California, Texas)

Medium Risk

Scope: SaaS companies with users in regulated US states

Colorado AI Act: developer transparency to deployers for high-risk AI. California SB 942: AI content labeling. Texas AI law: bias testing requirements.

Deadline: 1 Jan 2027 (Colorado, SB 26-189)

EU Product Liability Directive (revised) — covers AI software

Medium Risk

Scope: AI providers and operators placing AI-embedded products in EU

Strict liability for defective AI software causing harm, easier claimant access to evidence, presumption of defectiveness/causality in specified circumstances

Deadline: December 2026 (in force; the separate AI Liability Directive proposal was withdrawn in 2025)

What to do first

1

Classify your AI products: are they high-risk? Check EU AI Act Annex III against every product

2

GPAI model providers: publish transparency documentation now, regardless of scale

3

If you build high-risk AI: complete technical documentation and prepare conformity assessment

4

If you use high-risk AI (deployer): get documentation from your AI vendor, they must provide it

5

Data processor agreements with all AI vendors that process personal data

6

Colorado AI Act (SB 26-189, effective 1 Jan 2027): if you sell AI to deployers, you need developer transparency documentation

Estimated compliance cost

$15,000–$80,000 initial (varies widely by product type) + $5,000–$20,000/year ongoing

Proactive compliance typically costs 3–5× less than post-enforcement remediation.

Generate your technology AI compliance plan

Aegis Firma maps your specific AI systems against all applicable regulations for technology — and generates prioritized documentation across 169 jurisdictions.

Get Technology compliance plan

Further reading

Other industry guides