Language educators employ visual depictions of people for reasons including use as writing or conversation prompts, use as illustrations of vocabulary, or simply use as decoration. Despite research documenting so-called visual agism across mass media, there has been little attention to the issue in such language education materials. This project responded by creating and testing a framework for the detection and redress of agism in language teaching visuals. Drawing upon critical media studies, these efforts centred upon detecting attenuated numerical representation and stereotypical representation of older people in visuals. Given concerns with potential AI bias in social group representation coupled with the rapid uptake of AI in education, this framework was then tested by using it to evaluate visual materials created by generative AI tools intended for educational purposes. The results indicated the presence of bias against older people in these AI-generated images, suggesting a new avenue for the propagation of agism. AI prompts were then tailored to avert the generation of ageist visuals. This exercise thus empirically confirmed the value of the agism detection and response tools developed, demonstrating their usefulness for language educators and highlighting the importance of critical visual literacy among language educators and also among students.