The dual-use nature of AI, serving both legitimate and malicious purposes, poses a significant challenge for ethical hackers. They must navigate a complex ethical landscape, balancing the need to counter rapidly evolving AI-driven threats with the responsibility to uphold professional ethics, often blurring the line between ethical and unethical practices. The study explores how AI shapes ethical hacking practices. Using a qualitative approach, we analyzed interview and digital trace data to uncover five key practices where tensions emerge: augmenting hacking capabilities, interrogating AI, managing risk, bypassing ethical restrictions, and adapting AI for emerging threats. These tensions reflect trade-offs between ethical principles such as autonomy, beneficence, non-maleficence, explicability, and justice. The findings contribute to the cybersecurity literature on ethical hacking.