Deepfake phishing attempts use artificial intelligence (AI) generated synthetic fake content to trick users into revealing private information. These sophisticated threats are difficult for traditional cybersecurity methods to identify. In order to identify and stop deepfake phishing, this research suggests a GAN fingerprinting technique. We enhance the accuracy and traceability of detection by incorporating synthetic fingerprints into phishing content produced by GANs. To distinguish between genuine and fake phishing attempts, we use fingerprinted datasets to build a convolutional neural network (CNN) model. High detection accuracy and resilience to hostile attacks are demonstrated by the experimental results. A proactive approach to reducing the hazards of deepfake phishing is provided by the suggested solution.