The answer to this is a bit complicated as goseq depends on a few different databases. For the gene length information it will fetch data from either the geneLenDataBase package or TxDB annotation package or download from UCSC on the fly. GeneLenDataBase is not updated, so the Ensembl release used there will be old. Goseq also uses the org annotation packages (e.g. org.Hs.eg.db) to get the mapping between genes and go terms. I'm not sure how frequently the Ensembl versions are updated for that. If you want to ensure the latest version, you could use biomart or other sources (e.g. lengths from featureCounts or other summarisation tool) to get the length and GO mappings and specify these manually to goseq.
In case you need the lengths for Ensembl genes; the EnsDb.Hsapiens.v75 package provides Ensembl gene annotations for release 75, so it's not the newest, but it should be easy enough to generate EnsDb databases/packages for other, newer, Ensembl releases using their GTF files (check the ensembldb package).
library(ensembldb)
library(EnsDb.Hsapiens.v75)
## Get the gene lengths
geneL <- lengthOf(EnsDb.Hsapiens.v75, of="gene")
## Or of transcripts
txL <- lengthOf(EnsDb.Hsapiens.v75, of="tx")