In　this　letter,　we　propose　an　equality-constrained　detector　using　probabilistic　approaches　for　massive　multiple-input　multiple-output　(MIMO)　systems　with　generalized　space　shift　keying　(GSSK)　modulation.　Probabilistic　approaches　derived　from　the　Bayesian　iteration　can　provide　the　soft　decision　for　achieving　the　near-optimal　performance　and　some　other　benefits　such　as　combination　of　detection　and　channel　decoding.　However,　these　methods　cannot　be　directly　adopted　to　GSSK　modulated　MIMO　systems　without　considering　the　equality-constraint　of　GSSK　signals.　To　address　this　problem,　we　develop　an　augmented　system　model　with　a　weighting　parameter.　Moreover,　the　relation　between　the　augmented　model　and　the　quadratic　penalty　function　is　also　derived　to　determine　the　value　of　weighting　parameter.　Consequently,　probabilistic　approaches　based　on　the　augmented　system　model　can　be　directly　used　to　solve　the　equality-constrained　problems.　Simulation　results　substantiate　the　high-performance　and　extensibility　of　the　proposed　detector.