Advances in artificial intelligence (AI) technology, particularly deepfake technology, have increased the risk of hoaxes and disinformation spreading on social media, making it necessary to improve the public’s digital literacy. This community service project aims to enhance the ability of homemakers to detect AI-generated hoax videos in Sidodadi Village, Bengkulu Tengah Regency. The activity was conducted using the Participatory Action Research (PAR) approach, which includes the stages of To Know, To Understand, To Plan, To Act, and To Change. The target participants were 20 housewives who are active social media users. The program began with observations and Focus Group Discussions (FGDs) to identify digital literacy issues, followed by the development of training materials, interactive training sessions, hands-on practice in identifying AI-generated (deepfake) videos, and an evaluation using pre-test and post-test methods. The results showed that the training successfully improved participants’ skills across all digital literacy indicators. Before the training, participants’ average proficiency scores ranged from 3.2 to 3.7 (moderate category), while after the training, they increased to 4.4–4.6 (high category). The greatest improvements were observed in the ability to recognize the characteristics of AI-generated videos, distinguish between authentic and manipulated videos, verify information sources, and exercise greater caution before sharing content on social media. In addition to enhancing knowledge, this activity also fostered participants’ critical thinking when evaluating digital information. This community service program demonstrates that needs-based digital literacy training effectively strengthens community resilience against the spread of AI-based hoaxes and encourages homemakers to become agents of digital literacy within their families and communities.