AA Tau was observed for about 5h per XMM orbit (2 days) over 8 successive orbits, which covers two optical eclipse periods (8.2 days). The XMM optical/UV monitor simultaneously provided UV photometry with a ~15 min sampling rate. Some V-band photometry was also obtained from the ground during this period in order to determine the dates of the eclipses. Two X-ray and UV measurements were secured close to the center of the eclipse. The UV flux is the highest just before the eclipse starts and the lowest towards the end of it. We model the UV flux variations with a weekly modulation (inner disk eclipse), plus a daily modulation, which suggests a non-steady accretion. No eclipses are detected in X-rays. For one measurement, the X-ray count rate was nearly 50 times stronger than the minimum observed level, and the plasma temperature reached 60 MK, i.e., a factor of 2-3 higher than in the other observations. This X-ray event, observed close to the center of the optical eclipse, is interpreted as an X-ray flare. We identify the variable column density with the low-density accretion funnel flows blanketing the magnetosphere. The lack of X-ray eclipses indicates that X-ray emitting regions are located at high latitudes. Furthermore, the occurrence of a strong X-ray flare near the center of the optical eclipse suggests that the magnetically active areas are closely associated with the base of the high-density accretion funnel flow. We speculate that the impact of this free falling accretion flow onto the strong magnetic field of the stellar corona may boost the X-ray emission (abridged).