AI is often designed to replace human decisions. The authors argue that focusing too much on machine independence can lead to bias, ethical problems and lost trust, and can even make organisations worse at spotting security threats. This paper proposes AI-TMF, a framework that extends a leading US AI risk standard by adding the human side: users’ values, attitudes and business goals. Alongside technical checks, it uses workshops, conversations with users and behavioural approaches to assess social risks. It also covers fairness, accuracy and the impact of AI decisions. The goal is AI decision-support that people find transparent, fair and worth trusting.