<![CDATA[This explanatory sequential mixed-methods study examined Vietnamese EFL teachers' readiness and challenges in integrating AI-generated feedback into writing instruction. Quantitative data from 233 university teachers revealed moderate overall readiness (M = 3.09, SD = 0.58) masking pronounced disparities: ethical awareness (M = 3.76) substantially exceeded technological (M = 2.87) and pedagogical readiness (M = 2.54). AI usage emerged as the dominant predictor (β = .58, p < .001), with users demonstrating significantly higher readiness across all dimensions (d = 1.20-1.43) than non-users, while demographic variables showed null effects. Teachers reported high concerns about student over-reliance (M = 4.21), training deficits (76.4% received none), and pedagogical uncertainty (M = 4.08). Qualitative interviews with 15 teachers exposed pedagogical disorientation, ethical dilemmas without institutional frameworks, technical skill deficits masked by surface familiarity, and institutional abandonment. Despite challenges, successful adopters developed pragmatic strategies including division of labor, feedback sequencing protocols, and critical evaluation integration. Findings reveal "aware incompetence" wherein ethical consciousness outpaces implementation capabilities, functioning as a barrier rather than facilitator. Results challenge deficit narratives about teacher resistance while exposing how institutional policy vacuums (69.5% lacked policies) impede integration. Implications emphasize reconceptualizing professional development beyond technical training to address pedagogical frameworks, ethical protocols, and ongoing implementation support.]]>
