AI systems should perform consistently throughout their lifecycle and meet an appropriate level of accuracy, robustness and cybersecurity, in accordance with the generally acknowledged state of the art. The level and the relevant accuracy and robustness metrics should be communicated to the deployer. Accuracy relates to the degree to which AI system outputs correctly reflect reality. Robustness addresses how well an AI system maintains consistent performance despite variability in its operational conditions or environmental circumstances. Cybersecurity ensures that the AI system is protected from exploitation of vulnerabilities. The level of accuracy and accuracy metrics of the AI system should be established based on its intended purpose. Technical robustness is a key requirement for high-risk AI systems. They should be resilient in relation to risks connected to the limitations of the system in terms of accuracy and potential erroneous outputs and in relation to risks arising from actions by malicious third parties or other adversarial conditions created for the purpose of exploiting potential system vulnerabilities.