Clause 5.3: Functional adaptability
Functional adaptability describes how far the system can adapt to a changing environment, including by taking in fresh training data, live production data and what earlier actions produced.
What clause 5.3 is
Functional adaptability describes how far the system can adapt to a changing environment, including by taking in fresh training data, live production data and what earlier actions produced. Continuous learning is one route to it, not a condition. The model also warns that adaptability does not improve on a simple scale: reinforcing uncertain decision paths can make unwanted outcomes more likely and can entrench human bias.
What the 2023 edition said
The first edition added the characteristic under functional suitability, with the same warning about non-linear effects and bias reinforcement.
What the second edition changes
No change of substance is recorded in the public draft. The characteristic keeps its name and parent.
The FDIS text is not public. The change line above is read from the public draft, which is not the FDIS and may differ. No clause level diff is invented here. The ISO stage record is the authority for the dates.
Why it matters to the buyer
Adaptability is the quality that makes an AI system useful and hard to assure at the same time. The model refuses to treat more adaptation as simply better, which is the sentence to use when someone proposes continuous learning without monitoring.
What to write in the requirement
State what the system may learn from, how often, and what triggers a re-evaluation. Set a boundary on the behaviour that may change without approval. Monitor for drift and for the reinforcement of the decision paths the model warns about.
Evidence a reviewer asks for
A reviewer will ask what the system is allowed to learn and what would stop it. Continuous learning with no boundary is a finding.
- Adaptation scopeThe data sources the system may learn from, the update frequency and the approval gate
- Change recordEach model or behaviour change, with the date, the trigger and the re-evaluation result
- MonitoringDrift and outcome monitoring that would catch a harmful adaptation
Where it commonly fails
- The model retrains with no approval gate
- Adaptation is claimed but not monitored
- The evaluation set is older than the model
What to do before publication
Write the adaptation boundary and the retraining gate before the next model update, not after it.
Read across
Source for clause 5.3: ISO/IEC 25059, Clause 5.3, retrieved 20 October 2026. The change line rests on the public draft of the second edition and ISO page 88234. The requirement, evidence and gaps are written from our record. Every source.