UV absorption cross sections are an essential ingredient of photochemical atmosphere models. Exoplanet searches have unveiled a large population of short-period objects with hot atmospheres, very different from what we find in our solar system. Transiting exoplanets whose atmospheres can now be studied by transit spectroscopy receive extremely strong UV fluxes and have typical temperatures ranging from 400 to 2500 K. At these temperatures, UV photolysis cross section data are severely lacking. Aims. Our goal is to provide high-temperature absorption cross sections and their temperature dependency for important atmospheric compounds. This study is dedicated to CO2, which is observed and photodissociated in exoplanet atmospheres. We also investigate the influence of these new data on the photochemistry of some exoplanets. We performed these measurements for the 115 - 200 nm range at 300, 410, 480, and 550 K. In the 195 - 230 nm range, we worked at seven temperatures between 465 and 800 K. We implemented the measured cross section into a 1D photochemical model. For wavelengths > 170 nm, the wavelength dependence of ln(cross-section_CO2(wavelength, T)x1/Qv(T)) can be parametrized with a linear law. Thus, we can interpolate cross-section_CO2(wavelength, T) at any temperature between 300 and 800 K. Within the studied range of temperature, the CO2 cross section can vary by more than two orders of magnitude. This, in particular, makes the absorption of CO2 significant up to wavelengths as high as 230 nm. The absorption cross section of CO2 is very sensitive to temperature. The model predicts that accounting for this temperature dependency of CO2 cross section can affect the computed abundances of NH3, CO2, and CO by one order of magnitude in the atmospheres of hot Jupiter and hot Neptune. This effect will be more important in hot CO2-dominated atmospheres.