The basic concept might still be salvaged. I'll elaborate on my previous suggestion:
A description of the item might be encoded in such a way so as to avoid directly dealing with the object; this set of clues would then be handed to someone with no experience with SCP-586. This person would then be asked to assemble the description in especially vague language ("any object with the ability to", "might take the form of a cylinder" ). Special care would have to be taken to make sure the researcher taking part in this did not know that such an object existed. Finally, this output would then be handed to a third person, again with no knowledge of SCP-586, who would be asked to compile it into a full-fledged, proper, professional report, perhaps under the pretence of a test or practice for writing other reports.
Well, I think that would be enough to slip under the object's radar, so to speak. Chaining the process of writing through people who don't know the object exists, without ever referring to the object itself, might be able to bypass this verbal interference. Perhaps a report could be written "using" this method, which would be described in the article? The old report could be given joke status or archived as an example of 586's effect.