In general, it seems like GOF research doesn't have a crisp definition or a transparent reporting process. If it were possible to come up with criteria for evaluating whether or not a particular publication constituted GOF research and to easily identify studies that met those criteria, it would be possible to look at whether publication of such research has been increasing over the last 5 years, and use that as a proxy for the amount of research being done. But I don't think that's currently possible.

The NSABB is chaired by Samuel Stanley, who quoted in 2017 "I believe nature is the ultimate bioterrorist and we need to do all we can to stay one step ahead.” So it seems like NSABB is likely to remain an opaque institution, being run as a sort of security theater to justify continued GOF research while avoiding scrutiny.

"We saw in Chapter 7 the estimates of the rate of a laboratory acquired infection (LAI) range from 0.000057 to 0.001 per person per year, and the risk of onward transmission ranges from 0.05 to 0.15. If assume that 200 people are working on gain of function experiments in virology involving PPPs, and we place the risk of LAI rate conservatively at 0.0001 per person per year, then this would yield a risk of an LAI of 0.02, or one LAI every fifty years. If we assume the rate of onward transmission of a LAI is 0.15, then this would give us an annual risk a LAI leading to community infection of 0.003, or about one event every three hundred years.

Of course, this estimate of the risk depends on several assumptions we have made, and different people might arrive at different estimates. A great deal depends on how many people are working on this research and their adherence to biosafety protocols. Fouchier, one of the principal investigators who conducted to controversial H5N1 experiments, has estimated the risk of a LAI from this type of work to be 1 LAI per 70,000 persons per year or 0.000014 (0.0014%) per person per year (Fouchier 2015). Fouchier’s risk estimate was based on LAI data for people working in BSL 3 labs. Fouchier estimated the risk of onward transmission of a LAI to be between 0.00000025 and 0.00003 for a probability of onward transmission from a LAI related to this type of work of 1 person per 33 billion years (Fouchier 2015). Other researchers have calculated the risk of onward transmission to be much greater than Fouchier’s estimate. Klotz and Sylvester (2012) estimate the risk of onward transmission of a LAI from gain of function experiments in virology to be 0.003 per laboratory per year or once per 536 years. However, they also factor in the number of labs that may be conducting this research (42) for an estimate of one event every 12.8 years. Lipsitch and Bloom (2012) and Lipsitch and Galvani (2014) also consider the risk of onward transmission of a LAI from this type of research to be much higher than Fouchier’s estimate."

Dual Use Research in the Biomedical Sciences, 2021
 

The NSABB reviews GOF research in the USA, but:

"A committee at the Department of Health and Human Services is now tasked with making final decisions about what is and is not a gain of function experiment. But longtime critics of federal oversight still aren't pleased. Ebright claims that apart from two projects carried over from prior to 2017­ — the Fouchier and Kawaoka research on H5N1 — there are no public records of any reviews being performed, suggesting that "the NIH doesn't flag projects for review, nullifying the policy.

Arturo Casadevall, a microbiologist at the Johns Hopkins Bloomberg School of Public Health, co-authored a 2020 editorial in the journal mBIO, reporting that both PC30 reviews to date have been made behind closed doors, despite guidance from the White House at the time the framework was released encouraging a transparent process."

While I have not tried to measure whether search term frequency is closely correlated with the amount of interest or activity in a field of research, I often find it useful, and believe that the search term growth rate likely corresponds somewhat to change in funding and/or research activity and its offshoots. 

Below, I have searched three terms (with quotes) on Google Scholar, not including citations, for years in the close interval [2000, 2021]. Note that (1) corresponds to cumulative results, and (2) corresponds to results for that particular year.

I am unsure how to resolve the massive differences between the number of search results for "gain-of-function" and "gain-of-function research". From skimming the articles that appeared after each search, one explanation might be that there is indeed a lot of gain-of-function research that only uses the term gain-of-function (e.g., "gain-of-function mutation"). I still don't suspect the overwhelming majority of the results from searching "gain-of-function" actually refer to gain-of-function research, but the evidence that many of the results are What We Want leads me to believe that this graph is probably useful for looking at the growth in the field.