Compliance
EU AI Act Implementation
StatusImplementation statement
ScopeApplied AI systems and agents
A5 Tech builds applied AI tools, agents and research software. We classify each engagement by intended use, risk, user impact, data sensitivity and whether the system could fall under a regulated AI Act category.
AIA.01Use-case classificationBefore build, we record intended purpose, users, affected persons, data categories, model dependencies and whether the system is prohibited, high-risk, limited-risk or general-purpose dependent.Before scope
AIA.02No prohibited useWe do not knowingly build manipulative, exploitative, unlawful biometric, social scoring or other prohibited AI systems.Refusal rule
AIA.03TransparencyWhere users interact with an AI system or AI-generated output, the interface and documentation should make that clear unless an exception applies.User notice
AIA.04Human oversightSystems that affect people materially are designed for human review, contestability and traceable decisions rather than autonomous blind acceptance.Human in loop
AIA.05Evaluation and logsEvery delivery should include measurable acceptance criteria, known limits, model/version notes and, where appropriate, logging for monitoring and incident review.Eval harness
AIA.06Client responsibilitiesThe client controls deployment context, users, data and downstream decisions unless otherwise agreed. Orders should assign provider/deployer responsibilities clearly.Order terms