AI as a Service (AIaaS)

AI as a Service (AIaaS) refers to cloud-based artificial intelligence capabilities provided on a subscription or pay-per-use basis, enabling organizations to leverage AI without developing, training, or maintaining their own AI systems. Major providers include AWS, Azure, Google Cloud, and specialized AI companies.

Service Models:

• Pre-built AI services: Ready-to-use APIs for vision, language, speech (e.g., AWS Rekognition, Azure Cognitive Services)
• ML platforms: Infrastructure for custom model development (e.g., SageMaker, Vertex AI)
• Foundation model APIs: Access to large language models (e.g., OpenAI API, Claude API)
• AutoML: Automated machine learning for non-experts

EU AI Act Implications:

• Provider status: AIaaS providers may be "providers" under Article 3(3) for their services
• Shared responsibility: Both AIaaS provider and customer/deployer have obligations
• GPAI rules: Foundation model APIs subject to Chapter V requirements
• High-risk classification: Use case determines risk level, not service itself

Regulatory Challenges:

• Determining responsibility between provider and customer
• Ensuring compliance when AI is composed from multiple services
• Transparency about model capabilities and limitations
• Data protection for training and inference data

Related concepts: General-Purpose AI Model, Provider, Deployer, AI Value Chain