In trying to answer the question of whether nuclear risk has risen or fallen in the last three decades, I think it is useful to distinguish between military or terrorist risk on one hand and civilian or commercial risk on the other.
In the case of military risk, understood here as conflict between two or more states, although the rising number of countries with proven (or strongly stiving to achieve) nuclear capabilities would suggest that the risk of a nuclear confrontation has been increasing and is likely to continue to increase, empirical evidence suggests otherwise. Even considering the rise and challenges posed by countries such as Iran or North Korea, I would think that the risk of a nuclear military conflict has not been increasing, probably due to the fear that each and every would-be nuclear attacker has of the nuclear response that would be suffered immediately after any such attack.
One could recall that the only time in history that nuclear weapons were used in a military conflict was at the end of WWII in which the US bombed Japan who, in turn, lacked the capability of a reciprocal nuclear response. Had Japan been able to respond, I believe the US may have not used nuclear weapons in the first place. This deterrent effect my, in fact, be getting stronger as more nations acquire nuclear capabilities.
On the other hand, I believe that the risk posed by a terrorist attack using nuclear weapons or (nuclear waste material) might be increasing to the extent that as more potential sources of weapons or waste material become available, the likelihood of a security failure allowing access to these nuclear resources is increasing, particularly if one consider that an increasing number of warheads or nuclear wast is located in countries that may lack first rate security controls and/or may be permissible to corruption or even sympathetic to certain terrorist groups or rogue states.
Of course, for each country that does not adhere to, or leaves, the NPT gloabally increases the nuclear risk we all face.
Taking all this into account I would suggest that in the last 30 years, the highest nuclear risk humanity has faced was felt in the aftermath of the disintegration of the former soviet union, which had the potential to create new nuclear states and where control of nuclear weapons and equipment was week.
Regarding civilian and commercial side of nuclear risk, I would contend that these are most likely due to either the production of nuclear energy gone wrong or a security failure related to the disposal or storage of nuclear waste.
On the one had, after a small decline in the previous years, the number of nuclear power plant is on the rise again. Pressure to deal with climate change will likely mean that this trend will continue. Although the sheer rise in the number of nuclear reactors, compounded by the overall aging of the existing plants, elevates the risk of a meltdown in any one of them, I would think that security improvements on new, but also on existing plants (including periodic safety inspections before licensing renewal), would help reduce the severity of any single power plant melt down. In fact, while both Chernobyl and Fukushima accidents were devastating on several levels, the death toll of the former is estimated to be in the range of (4,000 to 60,000) whereas the more recent accident in Fukushima is estimated to have caused under 600 deaths (source: https://ourworldindata.org/nuclear-energy)
It seams likely that the trend of better safety will continue in the near future and ofset the rising risks of the increase of the number of nuclear plants.
All considered, I think that in the last 3 decades, the global nuclear risk has been decreasing and is likely to continue during the next 3 decades, although probably at a lower pace.
Notwithstanding the above considerations, below are three predictions on Metaculus regarding questions related to this issue.
I think that North Korea has developed is missile program as a means of responding to what it perceives as an existential threat from mainly the US and South Korea. It also also served as an international remainder of its growing capabilities (including the suspected capability to launch nuclear warheads) that might be used if and when international peace/settlement talks come to fruition. As such, there may not be any incentive for the current regime to reduce its current rate of testing at roughly 6 per year. Since I think that the current political regime is likely to remain in power, my best prediction in the time frame of the question would be about 12 tests.
The NPT is a first step in insuring that nuclear capabilities do not disseminate all over the globe, potentially to unstable and rogue regimes capable of jeopardizing global stability and peace. As such, any withdraw from NPT would most likely be met with strong international pressure and probably economic sanctions. A such, at least initially, from the point of view of the potential NPT dissenter, it might make more sense to covertly violate the treaty (while trying to maintain plausible deniability) than to formally withdraw. I think that in the time frame of the question there sohuld be a very low probability (in the range of 0% to 5%) on any NPT withdraw.
India, Israel and Pakistan live in a permanent state of military awareness and have deep distrust of the regional rivals, some of which are nuclear powers.
I do not believe that Israel, being surrounded by what it perceives as irreconcilable enemies would "(...) further the goals of nuclear disarmament and general and complete disarmament" even if it had full coverage of the US military.
Pakistan would not abandon its military agenda unless - at the very least - India did so too, but India will not go this rout for fear of China and thus, neither India, nor Pakistan will likely join the NTP, at least in the time frame of this question. I would put the probability of at least one of these countries joining the NTP at almost 0%.