Thorium has tremendous promise as a future source of nuclear energy. It is the only material whose fissile product (uranium-233) produces sufficient neutrons in thermal fission to sustain its own consumption. The important implication is that a properly-designed thorium-fueled reactor, once started with fissile material, can continue to produce energy without additional fissile material.
Inherent in this capability is the assumption of chemical processing. This is most easily accomplished if the thorium is in the form of a fluoride salt dissolved in a mixture of other fluoride salts, as was proposed for the Molten-Salt Breeder Reactor by Oak Ridge National Laboratory in the 1960s. Several variants of the MSBR were proposed by ORNL, some with a single fluid containing both thorium and uranium-233 as dissolved fluorides, others with these two materials separated into a fuel mixture and blanket mixture. Chemical processing was needed in either scenario.
A letter to Nature  by Stephen Ashley (“Nuclear energy: Thorium fuel has risks“) claimed that the thorium fueled reactor had a danger due to the ability of protactinium to be chemically separated from thorium. Ashley claimed that protactinium separation would lead to the formation of pure uranium-233 which would be a material that would be suitable for a weapon.
Natural uranium contains 0.7% uranium-235, which has been made into nuclear weapons and used. Isotope separation techniques for uranium permit this material to be enriched, and this is a source of global controversy. The remaining uranium-238, exposed to neutrons in any sort of reactor, form plutonium-239 that has been made into nuclear weapons. Both of these paths are vastly simpler than any weapons program around uranium-233, which has never been developed into an operational nuclear weapon. Of the handful of nuclear weapons tests that have involved uranium-233, the results have been ambiguous. In 1951, at the zenith of American interest in nuclear weapons of any kind, military planners privately inquired if the US Atomic Energy Commission had any interest in developing thorium or uranium-233 weapons. Chairman Gordon Dean stated unequivocally that there were no plans to develop weapons using uranium-233. 
It is not difficult to understand why. Glenn Seaborg discovered uranium-233 at the University of California, Berkeley, in April 1941. Within a few months he was part of the Manhattan Project to develop nuclear weapons, specifically tasked with developing techniques for chemically separating plutonium, which he had previously co-discovered, from irradiated uranium in production reactors that were yet to be designed and built. The chemistry of plutonium was almost completely unknown, whereas the chemistry of uranium was clearly understood. By February 1942, his research had shown that uranium-233 was fissile like plutonium, and could be chemically separated from thorium just as plutonium could be separated from uranium. 
Seaborg was strongly incentivized to investigate the military uses of uranium-233 for a weapon, since chemical separation techniques could be perfected using natural uranium long before a production reactor was available. This was not the case with plutonium, where only micrograms of material existed and chemical separation research proceeded with exceptional difficulty. Seaborg pressed repeatedly on Manhattan Project leadership for permission to investigate uranium-233 for weapons use, and raised the spectre that the Germans might already be working on a U-233 weapon in order to bolster his argument. 
By the summer of 1944, it had become clear that any plutonium produced for the project would be heavily contaminated with plutonium-240 and unsuitable for a gun-type weapon. Seaborg and a team of reactor designers were given permission to investigate a way to fission the plutonium that would be produced in a special reactor and to generate uranium-233 to build the weapons for which they believed the plutonium would be unsuitable. 
The reactor design believed that such a reactor could be built, but Seaborg and his chemists soon found even more reason for dismay. They realized if the reactor was fueled with solid metallic plates, then each time these fuel elements were chemically dissolved for reprocessing, the rate of plutonium recovery would need to be nearly perfect, so they supported the idea of a reactor that would use a fluid fuel, where plutonium, thorium, and uranium were dissolved in water that would serve both as coolant and neutron moderator. This eliminated much of the concern about dissolution, purification, and refabrication of solid fuel elements, and even permitted the idea of protactinium isolation as it formed from neutron absorption in thorium. The fear that protactinium isolation from a fluid-fueled reactor would lead to the formation of uranium without any uranium-232 impurity was the subject of Stephen Ashley’s letter. 
What Seaborg found, even before completing a conceptual design, was that uranium-232 production could not be reduced sufficiently to produce an attractive fissile product for a weapon, even when accounting for the possibility of protactinium separation. On December 28, 1944, he sent a report to his leadership entitled “Conversion of Pu-239 to U-233” where he noted that there were two other pathways to the formation of uranium-232 in any thorium fueled reactor. Each pathway led to the formation of protactinium-231, whose neutron absorption was large and would form protactinium-232, which rapidly decayed to uranium-232. 
The first formation pathway was a fast-neutron reaction on thorium-232 that knocked out a neutron, forming thorium-231 which rapidly decayed to protactinium-231.
The second was neutron absorption in thorium-230, which was a tiny fraction of natural thorium present due to the decay of uranium-238. Neither formation pathway could be completely excluded from a practical reactor without causing unacceptable reduction in production rates. And the absorption cross-section of protactinium-231 was sufficiently high that even if protactinium was being actively removed, there would still be sufficient material present to form unacceptable levels of uranium-232 in the final product.
Seaborg himself petitioned for the termination of the uranium-233 investigation in early 1945  and later innovations in implosion-type weapons designs made the plutonium that had been produced acceptable for use in weapons, first in the Trinity test of July 1945 and later in the attack on Nagasaki.
What can be seen from this historical example is that even with the most brilliant minds in radiochemistry and the intense pressures of war upon them, uranium-233 from thorium was considered and rejected for weapons-related use. None of the underlying issues have changed appreciably since then.
- S. F. Ashley et al., “Nuclear energy: Thorium fuel has risks“, Nature 492, 31–33 (December 2012).
- “Atomic Shield, 1947-1952“, page 566.
- G. Seaborg, “Early History of Heavy Isotope Research at Berkeley,” Lawrence Berkeley Laboratory, University of California, Berkeley (1976), page 100.
- G. Seaborg, “History of Met Lab Section C-I: May 1943 to April 1944,” Lawrence Berkeley Laboratory, University of California, Berkeley (1978), pages 398-399.
- G. Seaborg, “History of Met Lab Section C-I: May 1944 to April 1945,” Lawrence Berkeley Laboratory, University of California, Berkeley (1979), page 352.
- A. Weinberg, “The First Nuclear Era”, pages 37-38.
- Kirk Sorensen, “Thorium Research in the Manhattan Project Era“, thesis, University of Tennessee, May 2014