AI-Generated Phishing Email Detection

Generated phishing and legitimate datasets using OpenAI and Anthropic APIs with varied prompts to ensure data diversity.

Developed preprocessing and vectorization methods to enhance machine learning input quality. Trained models including naïve Bayes, logistic regression, and random forest using sklearn and pandas.

Measured model accuracy with custom test datasets, iterated on feature selection, and optimized detection algorithms for real-world phishing scenarios.