I faced this kind of problem in a MBP tagged protein. Looks like the protein of interest has slight affinity and is partially bound to the impurities. They do not look like degradation products. It is not only lower sized proteins, there is also a band at very high molecular weight. One suggestion is to perform Size exclusion chromatography with higher salt content (500mM NaCl) in the buffer and maybe add 0.05% of detergent like TritonX100 if it does not affect your protein. This may help in de-coupling the impurities from the protein of interest.

GST has  a propensity to dimerize so I think what you are seeing is mixed full-length and truncated dimers of the GST fusion-this will make it difficult to purify by SEC. Unfortunately I don't know what the best way is to break the dimer-some papers suggest it is a hydrophobic interaction so detergents should help (as already suggested).

You could try to reduce the levels of truncated proteins by reducing induction temperature, using Rosetta or Codon plus cells, using a richer medium e.g. TB instead of LB  and/or using auto-induction.

Try to load the protein of interest into the Glutathione Sepharose column and then wash it with a high salt buffer before you elute the protein again. The high salt concentration helps to wash out most of the proteins, that have a low affinity to the column or to your protein of interest. For example I add additionally 400 mM KCl for this step. I tried both ways and with the additional high salt washing step I have higher purity of the protein.