Many disorders, mental and physical, are known to have genetic causes (sometimes requiring further environmental input to trigger susceptibility created by a genetic trait). Some of these genetic disorders have a cause in a single known gene or two in almost everyone who has the condition. This makes these diseases good candidates for gene therapy. In theory, one affixes the corrected DNA to a retrovirus, it infects the host and replaces the host's defective DNA with the corrected DNA of the virus, and the person is permanently cured of the condition.
But, many conditions, for example schizophrenia and bipolar disorders and autism, often (even if not always) have genetic causes. But, those genetic causes are massively polygenetic. A thousand different mutations cause symptoms that we understand to cause the disorder. Sometimes it takes just one of many different disease risk increasing diseases to cause the disorder, sometimes it takes many.
These conditions are much less susceptible to gene therapy because you have to figure out which particular genes cause the problem in each particular person and fix them all on an individual by individual basis. You need a custom cure for every patient with the immense amount of work that goes into devising a cure to any genetic disease at all, in every single case.
Suppose, however, that you are an evil, bad person, who wants to hurt a large number of people, or you think that in the interests of neurodiversity that some condition that everyone else thinks of as a defect is really a superpower of sorts that everyone should benefit from themselves. Then, the analysis is very different.
While there may be thousands of ways of causing a polygenetic condition, which makes it very difficult to treat with gene therapies, it takes a solution to only one of those thousands of ways to formula a gene therapeutic RNA virus to transfer the genes that impart susceptibility to condition to the person infected to infect millions upon millions of people in what could be self-sustaining if the virus gets out of control and starts to reproduce "in the wild."
Thus, it is much easier to make a polygenetic condition widespread than it is to make a cure for it. This insight came to me watching the television series "Terra Nova" (Episode 4, 2011) on Netflix, but this science fiction observation is a serious one. In the episode, a researcher seeking to cure a form of dementia with a simple genetic cause that he knows that he has due to genetic testing seeks to cure it with viral gene therapy, but rather than curing it, he mistakenly creates a gene therapy virus that changes the relevant gene in a way that produces the rapid onset of the condition rather than preventing it. The virus soon infects many people until our intrepid heroes reverse engineer a cure with a bit of lucky serendipity.
But, while accidents of the kind depicted could probably be easily contained in the real world, and would go through layers of animal pre-testing first that would catch the error, the same assurances wouldn't be present in the case of someone who wanted to intentionally cause specific individuals or members of general public to develop a genetic condition, good, bad or indifferent. Indeed, there are known "selective fitness neutral" retroviruses out there "in the wild" infecting people and changing their DNA makeup laterally right down to their germ lines today. Creating a virus that carries a well characterized harmful gene whose workings in humans are already well understood might be easy compared to other forms of gene therapy. Yet, it would also be a subtle attack which would be hard to distinguish from a natural lateral gene transfer without advanced scientific knowledge.
One could also imagine some subculture of people who intentionally tried to modify their own genomes with known genetic traits of others, where people are less careful than they should be about containing the virus of transmission of the genetic traits in a way that causes the genome changing virus to spread uncontrollably.
From an epidemiologist's point of view, particularly with a slow onset or subtle genetic condition imparted by a mild carrier virus with few other symptoms, the only real clue to the orgin of the threat would be the sudden appearance of a single genetically distinct subtype of what is usually a massively polygenetic condition marked by a distinct uptake in the rate of which the genetic condition appears. But, since people who are diagnosed with massively polygenetic conditions are rarely genotyped in the course of treatment for that condition, it could evade detection for a very long time, at which point the perpetrator might be virtually impossible to trace.
What would you do if one day you woke up and noticed that 10% of the population of Chicago had adult onset severe autism that had manifested sometime in the last two years? What if only people who went into the U.S. Capital building were infected but the rate was proportional to the time spent there and was nearly 80% for people who were there for two continuous years? What if the virus caused only a local infection and was transmitted via the lubricant in a single brand of condoms with a national distribution, leading to genetic change in the children of people who used the brand later when they stopped using contraceptives but was invisible in a DNA test drawn from siliva or hair from the scalp, for example? The potential for massive permanent harm is great indeed.