A security team uses an AI system to analyze log data for signs of compromise. The AI was trained on labeled datasets from 2021-2023. A new threat group emerges in 2024 using entirely novel techniques not seen before. What detection limitation will the team face?

• A.The AI can detect the new techniques because adversarial training makes it inherently robust to novel attacks; adversaries cannot probe a deployed detector without triggering it, since every evasion attempt is by definition an anomaly the model flags
• B.The AI will classify all 2024 logs as malicious because the threat group's activity exceeds 2021-2023 baseline volumes; model monitoring is needed only for online-learning systems, since a frozen model's accuracy cannot change after deployment
• C.The AI has no training signal for the new techniques, so it will likely miss them or only flag them if they happen to resemble known-malicious statistical patterns from training data
• D.The AI cannot analyze log formats from 2024 because they differ from 2021-2023 formats