I would recommend investigating the results more closely and look for things like the accuracy of the classification, how many different bacteria/transcripts seem to be present

As a next step, if you were to extract and map the reads that were classified as that organism how well do those reads cover the expected "contaminants".

Can you recover reasonably full transcripts? or all reads are covering the same tiny fragment of the genome. How well do the reads match?

Most often what you will find are reads that only match partially, and only in a single tiny, region, and that would indicate artifacts of the sequencing process.