Ice binding proteins modulate ice nucleation/growth and have huge (bio)technological potential. There are few synthetic materials that reproduce their function, and rational design is challenging due to the outstanding questions about the mechanisms of ice binding, including whether ice binding is essential to reproduce all their macroscopic properties. Here we report that nanoparticles obtained by polymerization-induced self-assembly (PISA) inhibit ice recrystallization (IRI) despite their constituent polymers having no apparent activity. Poly(ethylene glycol), poly(dimethylacrylamide), and poly(vinylpyrrolidone) coronas were all IRI-active when assembled into nanoparticles. Different core-forming blocks were also screened, revealing the core chemistry had no effect. These observations show ice binding domains are not essential for macroscopic IRI activity and suggest that the size, and crowding, of polymers may increase the IRI activity of "non-active" polymers. It was also discovered that poly(vinylpyrrolidone) particles had ice crystal shaping activity, indicating this polymer can engage ice crystal surfaces, even though on its own it does not show any appreciable ice recrystallization inhibition. Larger (vesicle) nanoparticles are shown to have higher ice recrystallization inhibition activity compared to smaller (sphere) particles, whereas ice nucleation activity was not found for any material. This shows that assembly into larger structures can increase IRI activity and that increasing the "size" of an IRI does not always lead to ice nucleation. This nanoparticle approach offers a platform toward ice-controlling soft materials and insight into how IRI activity scales with molecular size of additives.