Although AI technologies show great promise for education, their inclusion into assessment systems has generated debates regarding student motivation, anxiety, learning opportunities, and academic results. This study explored the influence of teacher support in AI-assisted exams on L2 learners' demotivation, anxiety, L2 learning experience, and academic success. Conducted at a large university in Ethiopia, participants included 92 BA Management students from two intact classes, equally distributed by gender and ranging in age from 18 to 23. The two intact classes were randomly divided into an experimental and a control group. Using a quantitative quasi-experimental pretest-posttest control group design, the study administered an Oxford Quick Placement Test, the Academic Motivation Scale, the Foreign Language Classroom Anxiety Scale, the L2 Learning Experience Scale, and a researcher-made test to assess academic success. The AI tools integrated into the exams included automated assessment and feedback systems to enhance learner engagement. Chi-square analyses and independent samples t-tests revealed significant positive effects of teacher support on reducing demotivation and anxiety, enhancing L2 learning experiences, and improving academic success in the experimental group compared to the control group, highlighting the benefits of combining AI tools with teacher support. These findings suggest that teacher support in AI-assisted exams can substantially benefit L2 learners. Additionally, the findings indicate that AI-assisted exams can considerably improve learning outcomes when paired with effective teacher involvement, highlighting implications for various stakeholders in L2 instruction. Implications of the study, limitations, and suggestions for future research are discussed.