The obliquity of a planet is the tilt between its equator and its orbital plane. Giant planets are expected to form with near-zero obliquities [1,2]. After its formation, some dynamical mechanism must therefore have tilted Saturn up to its current obliquity of 26.7 • . This event is traditionally thought to have happened more than 4 Gyrs ago during the late planetary migration [3,4,5] because of the crossing of a resonance between the spin-axis precession of Saturn and the nodal orbital precession mode of Neptune [6]. Here, we show that the fast tidal migration of Titan measured by [7] is incompatible with this scenario, and that it offers a new explanation for Saturn's current obliquity. A significant migration of Titan would prevent any early resonance, invalidating previous constraints on the late planetary migration set by the tilting of Saturn [8,9,10]. We propose instead that the resonance was encountered recently, about 1 Gyr ago, forcing Saturn's obliquity to increase from a small value (possibly less than 3 • ), up to its current state. This scenario suggests that Saturn's normalised polar moment of inertia lies between 0.224 and 0.237. Our findings bring out a new paradigm for the spin-axis evolution of Saturn, Jupiter [11], and possibly giant exoplanets in multiple systems, whereby obliquities are not settled once for all, but continuously evolve as a result of the migration of their satellites.We investigate whether the spin-axis dynamics of Saturn could have been influenced by the migration of its satellites. The torque applied by the sun on Saturn's equatorial bulge produces a precession of its spin axis with a mean frequency ψ = α (1−e 2 ) −3/2 cos ε, where e is the eccentricity of Saturn, ε is its obliquity, and α is called its precession constant [12]. The value of α incorporates the extra torques produced by Saturn's satellites [1,13]. It also depends on Saturn's orbital and physical parameters, among which all are well known except Saturn's normalised polar moment of inertia λ. Estimates found in the literature broadly range in λ ∈ [0.200, 0.240] but they are model-dependent and a proper uncertainty range is hard to define (see Methods). Exploring this whole interval gives a current precession constant α 0 ranging from 0.747 to 0.894 /yr.We take into account the migration of Saturn's satellites by changing their contributions to α. The measurements of [7] unveiled a fast migration for Titan, and a similar tidal timescale t tide = a/ ȧ for all six satellites of Saturn studied. Both these findings support the "resonance locking" tidal 1