📄 中文摘要
深度伪造检测器在市场宣传中常常声称具有96%的准确率，但在实际应用中，这一数字却降至65%。这种31个百分点的差距并非简单的校准错误，而是反映了我们在数字证据处理方法上的根本性缺陷。对于开发者而言，这一技术隐含的意义在于，概率检测在与深度伪造技术的竞争中处于劣势。实验室中训练的模型在真实环境中的表现远不如预期，这对生物识别系统和计算机视觉管道的开发提出了严峻挑战。
深度伪造检测器承诺96%的准确率，实际表现仅为65%
📄 English Summary
Deepfake Detectors Promise 96% Accuracy. In the Real World, They Drop to 65%.
Deepfake detectors often advertise a 96% accuracy rate, but in real-world applications, this figure drops to around 65%. This 31-point discrepancy is not merely a calibration issue; it signifies a fundamental flaw in our approach to digital evidence. For developers, the technical implication is that probabilistic detection is losing the arms race against deepfake technology. Models trained in controlled lab environments perform significantly worse in real-world scenarios, posing serious challenges for the development of biometric verification systems and computer vision pipelines.