[Tacit knowledge] In 1995, the Japanese cult Aum Shinrikyo unleashed terror on the Tokyo subway system with a highly publicized sarin gas attack. However, less is known about the group’s development of biological and chemical weapons and about their prior attacks using these weapons.

This CNAS report, Aum Shinrikyo: Insights Into How Terrorists Develop Biological and Chemical Weapons, culminates a multi-year project led by Richard Danzig, former Secretary of the Navy & Chairman of the CNAS Board of Directors; with Marc Sageman, Advisor to the Deputy Chief of Staff of the Army on the Insider Threat; Terrance Leighton, Senior Staff Scientist at Children’s Hospital Oakland Research Institute and Chief Scientist at Science Applications International Corporation; Lloyd Hough, Senior Research Scientist at Battelle in International Technology Assessments; Zachary Hosford, Research Associate at CNAS; and two Japanese colleagues investigating these issues. Through personal interviews and correspondence with former members of Aum Shinrikyo’s leadership, the report provides never-before documented information on the terrorist group and its operations. The observations from this study have wide-ranging applications for terrorist groups worldwide…Finally, we want to express our gratitude to an unusual source, past members of Aum Shinrikyo, some of whom have been sentenced to death for their roles in murders committed by the organization [since executed].

This detailed case study of Aum Shinrikyo (Aum) suggests several lessons for understanding attempts by other terrorist groups to acquire chemical or biological weapons. We provide the basis for these observations in the discussion that follows and return to them at greater length in the conclusion of this report.

…When the Sakamoto family’s disappearance was discovered, relatives saw that Sakamoto’s wallet was still intact and the apartment had not been robbed. They also noted that an Aum Shinrikyo badge had been found at the scene.⁵⁴ During this time, Shoko Asahara and several senior advisors, including some of the perpetrators, left Japan to visit a foreign office that Aum had earlier established in Bonn, Germany, so when the police came to interview him 16 days after the murder, he was not available.⁵⁵

The Sakamoto murders had characteristics that were to recur in the cult’s future biological and chemical attacks: an interchangeable approach to biological, chemical and conventional weapons; the clumsiness of much of the planning; the crude effectiveness, nonetheless, of the result; the absence of women from the planning group;⁵⁶ hands-on execution of the attack by generally senior members; containing knowledge among these members; only transitory moral scruples, if any, exhibited by the participants; responsiveness of members to their charismatic leader; and a delayed and easily distracted police follow-up.⁵⁷

The Sakamoto disappearance stoked press antagonism toward the cult.⁵⁸ Aum-bashing developed into a common pastime. Despite this, Asahara confidently predicted victory in the parliamentary elections. The result was disastrous. Asahara received only 1,783 votes out of half a million cast, and he appeared to have been shocked when none of the Shinrito candidates was elected.⁵⁹ The brief life of Shinrito was an unmitigated disaster for Aum. The campaign was extremely costly, about 7 million dollars, posing a serious short-term threat to the cult’s financial stability. Okazaki (one of the murderers of the Sakamotos) absconded with 1.5 million dollars. He later negotiated a deal with Asahara to keep a portion of this in return for his silence.⁶⁰ Many monks, who had left the communes to campaign, defected after exposure to the real world and returned to their former lives. The electoral rejection solidified Asahara’s belief that the people at large were not worthy of salvation.

…At the same time, Aum experienced substantial internal changes. In the late 1980s, Asahara had been surrounded by strong women, such as Tomoko Ishii (his wife), Hisako Ishii (his mistress), Iida and Sanae Ouchi. These women had been among his earliest disciples and had collectively exerted a powerful influence. By 1992, perhaps simply for personal reasons,⁸⁸ it seems that this group had faded away from top decision-making.

The turn toward violence both assisted, and was assisted by, the rise of a male group around Murai that replaced these female advisors. It appears that most or all of these men had been science fiction enthusiasts in their youth and continued this interest as adults. The group was consumed with the idea that spiritual experiences could be verified, explained, tested and proved scientifically. Projects included creating an astral teleporter (an electronic device to clean one’s “astral dimension”) and the “Perfect Salvation Initiation” headgear (designed to transmit Asahara’s brainwaves to his disciples and replace their own impure brain waves).⁸⁹ The group’s interest in science also applied to the violent aspects of the cult. They talked about futuristic technologies of destruction, such as plasma weapons that could atomize human bodies, mirrors several miles across that would float in space reflecting the sun’s rays so that they destroyed all life in the process, vast laser guns and other imagined means of destruction. It was presumed that many such weapons had been developed by the United States.

Asahara shared his scientists’ fascination with pop science. Although blind, he would turn on television animation shows and have someone describe the scenes. Many of his prophecies and sermons were taken from these shows. A plan for an underwater city after World War III came from Future Boy Conan. Light Beam Satellite Cannon was taken from Mobile Suit Gundam, and Cosmo Cleaner from Starblazers. His talk of Freemasons was taken from the magazine Mu [monthly occult/paranormal fanzine founded 1979]. One person testified, “Since we all knew the sources, we thought Asahara’s ideas to be foolish.”⁹⁰

…each fermenter might be expected to yield about 9,000 liters of material. Nakagawa recalled that 2–3 of these “fermenters” were constructed and that from them ~50 batches were made.⁷³ We could find no record or recollection of the precise amounts of material that resulted, but a simple calculation confirms the recollection of participants that these amounts were huge. The yield from one production cycle of one fermenter (9,000 liters) would equal 9 metric tons and 50 cycles would yield the immense amount of 450 metric tons. However, this quantity should not be equated with weapons material. The product, a yellow liquid, would have consisted mostly of the media in which the C. botulinum was cultivated. The cult made no efforts to concentrate or purify the broth from the fermenters. All indications are that the product, though voluminous, was not efficacious. Cult members recognized that no one died in their attacks. Indeed, when a member slipped and fell into a fermenting tank he nearly drowned, but did not show signs of disease.⁷⁴

…These issues notwithstanding, the cult employed 3 trucks⁷⁹ to spray this material at two US Naval bases, Narita airport, the Japanese Diet, the Imperial Palace and the headquarters of a rival religious group.⁸⁰ At the same time, Asahara proclaimed that those who were not members of Aum could not be saved and would inevitably descend into hell. He called on Aum members to flee from the coming disaster and save themselves by attending a 3-day seminar to be held on the island of Ishigaki, in the Ryukyu archipelago. He said that this would be a very important seminar and that all who could possibly attend should do so.

When the botulinum attacks failed, Aum leaders variously attributed this failure to a lack of virulence in the pathogen, recurring problems that interfered with the spraying equipment and poor prediction of shifting winds. Fortunately for Aum, these attacks went unnoticed

…Using tons of material from this production process,⁹⁹ Aum undertook further attacks with C. botulinum. In November 1993, and in the fall of 1994, two individual assassinations of perceived enemies of Aum were apparently attempted, the first from a sprayer and the second by concocting a mixture with juice.¹⁰⁰ Finally, 5 days before the cult’s fatal use of sarin on the Tokyo subway system in March 1995, the cult dispersed botulinum in the Kasumigaseki Subway Station.¹⁰¹… for example, the November 1993 assassination attempt of Daisaku Ikeda was apparently undertaken with both sarin and botulinum—the redundancy arising, in the view of contemporaries, because Endo wanted a basis for claiming at least a share of any credit.

…The more interesting question is why Aum employed a benign strain. Speculation has included hypotheses that Aum intended its anthrax dissemination only as a trial run, that someone sabotaged the effort or that an error was made when a benign rather than potent strain was obtained and that the error was never recognized or was ignored. Joyu and Nakagawa separately provided a different explanation. They were clear that all involved in this effort were well aware that they had a vaccine strain, but they believed that Endo could use “genetic engineering” (Joyu’s statement of Endo’s phrase) to convert this benign anthrax into a lethal form.¹¹² None of our informants claimed to understand Endo’s methodology. Without access to Endo—who declined our requests for an interview—we cannot assure the accuracy of any explanation as to how he intended to effect this conversion, what his chances of success were and why he failed.

We have developed a hypothesis about the path that Endo may have been pursuing,¹¹³ but emphasize that the supporting evidence is fragmentary. If we are not correct, the conversion may merely illustrate of what Endo could have had in mind. By this hypothesis, Endo obtained not only the Sterne strain but also the second vaccine strain—the Pasteur strain—mentioned by Makino as obtainable at Obihiro University.¹¹⁴ The Sterne strain has the first, but lacks the second, of the two plasmids (pXO1 & pXO2, respectively) that are together required to produce an effectively virulent anthrax bacterium.¹¹⁵ Endo may have realized that, conversely, the second (Pasteur) vaccine strain provides immunity while avoiding lethality by retaining the second plasmid and eliminating the first. Based on a 1989 publication by Russian scientists, Endo may also have been aware that if both benign strains were obtained, they could together be used to generate a virile strain containing both plasmids.

…Keim’s analyses of the extant Aum sample show that the strain that was sprayed lacked the second plasmid. Accordingly, it appears that whatever path Endo pursued, he failed. This conclusion is supported as well by Aum’s evident failure to kill anyone with its anthrax preparations…Makino recalled that when he first started his capsule research program a skilled co-worker could not successfully use the Pasteur strain as a genetic transduction donor of pXO2. Before he could successfully employ the plasmid transfer method, the co-worker had to spend 6 months in a leading laboratory at the Pasteur Institute to learn the nuances of the virus-mediated transduction methods. Even then, Makino described the transduction procedure as “very inefficient.”

…Both Nakagawa and Hayakawa separately told us that they suspected that Endo’s anthrax was not potent when they and others inadvertently inhaled some and did not fall ill.¹²⁴ Others developed similar doubts when anthrax attacks were undertaken and failed, particularly in June & July 1993.¹²⁷

…This unbroken string of failures with botulinum and anthrax eventually convinced the group that making biological weapons was more difficult than Endo was acknowledging. Asahara speculated that American comments on the risk of biological weapons were intended to delude would-be terrorists into pursuing this path. [Interview with Fumihiro Joyu (21 April 2008).]

…[Sarin] work proceeded around the clock. By September 1993, the production facility at Satyan 7 was declared ready for occupancy.¹⁵² However, this readiness appears to have been overstated. Perhaps because of the haste with which it was built, Satyan 7 never came close to the stated goal of 70 tons of sarin. As described below, it was capable of producing some 40–50 liters (that is, ~100 pounds) of the chemical. It eventually employed 100 Aum members and was equipped with 30-liter flasks with mixing and temperature control capabilities within enclosed protective hoods.¹⁵³ The photograph on the right conveys a sense of Satyan 7’s size. A subsequent United Nations report estimated that the building and its contents cost 30 million dollars.¹⁵⁴

…Despite the available evidence of Aum’s involvement with sarin production, police could not prosecute Aum without tying it to a particular murder, assault or kidnapping as no Japanese law prohibited manufacture of a poisonous gas.

…Our analysis suggests that the cult’s 1995 Tokyo subway attack would have been much more lethal if Aum had not destroyed its purer sarin when it feared discovery a few months earlier or disseminated the low-purity sarin more effectively…The cult’s chemical capability, however, has been understated because of the fortuity that it had to resort to low-purity sarin in its hastily arranged subway attacks. This reduced the number of casualties and diminished recognition of what Aum could achieve. Moreover, although Russian information may have aided planning the factory production system, we see no evidence of a Russian contribution to the cult’s method of sarin production. For the most part, the chemical program was both independently developed and efficacious. Aum’s activities illustrate how readily a terrorist group can produce chemical agents when operating under permissive conditions.

Why was Aum more successful at developing chemical than biological weapons? Many have observed that Tsuchiya was impressively skilled as a chemist and Endo unimpressively skilled as a microbiologist.¹⁹² Indeed, Endo’s training, interrupted by his joining Aum, was as a virologist not as a bacteriologist, while in Aum’s weapons program he worked with bacteria. The disparity in relevant skills undoubtedly affected the success of their respective programs. But we think that more pervasive, less serendipitous factors shaped Aum’s greater success at chemistry than at biology.

First, Aum’s history suggests that with access to the relevant literature, a generally skilled chemist can produce a chemical weapon, whereas a generally trained biologist is likely to have more difficulty propagating and conserving an unfamiliar pathogen. The distinction between explicit (book) knowledge and tacit (hands-on) knowledge may be helpful here.¹⁹³

Developing biological weapons appears to require more tacit knowledge, while chemists may be adequately positioned to develop weapons after consulting relevant documentation. Although he made some improvements and innovations of his own, Tsuchiya mastered the basics of sarin production and a number of other hazardous or illicit chemical compounds primarily through reading scientific journals. He was able to communicate recipes that Murai and others could employ with reasonable efficiency (though they made some errors and had accidents). By contrast, while Endo’s efforts to identify, harvest and grow toxigenic C. botulinum and to create fully virulent B. anthracis also relied on library research, the work evidently required more expert training and experience. Moreover, moving from one pathogen to another required essentially restarting the program. Even if Endo had developed the requisite skills, it would have taken him (or anyone else) considerable effort to pass them to others. Producing biological materials is a modern craft or an art analogous to playing a sport or speaking a language. Though some aspects can be mastered just from reading a book, others relevant to a weapons program cannot be acquired this way with rapidity or assurance.

…A third factor that made chemical weapons easier to develop was that chemicals of the required purity were readily accessible at low visibility for plausibly legitimate business purposes. By contrast, Aum’s efforts to harvest C. botulinum from the natural environment were unproductive—finding natural samples of the requisite and demonstrable potency is difficult.¹⁹⁶ The cult’s hospital work provided a rationale for purchasing biological materials, and other groups or individuals with terrorist intentions may use similar facades. But even in the 1990s, it was much more difficult for Aum to acquire virulent strains than it was to acquire chemical reagents and it apparently wanted to avoid the visibility that ordering these anthrax strains might have conferred. At present regulatory and policing constraints are strong. Although synthetic biology and other developments enhance opportunities for biotoxin synthesis, creating or procuring seed stock seems likely to be more difficult than procuring chemicals. (Obtaining nuclear materials and equipment would be harder still.)

Fourth, the potency and purity of chemical material can be assayed within an hour, and its killing effect thereby readily calculated. Biological material is more difficult to assess. Although by and large Aum understood its chemical weaponry, it never confidently or accurately knew what it had, or lacked, in the potency of its biological weapons. Reliably assessing of biological material will typically require extensive testing, often with animals. Moreover, pathogens—which are living organisms—are likely to evolve and/or degrade over time and under different conditions. In addition to posing challenges in storage, this amplifies difficulties in assessment. To this day neither we nor the leaders of Aum Shinrikyo know whether Endo possessed a fully virulent strain of B. anthracis and was unable to conserve it, or whether he conserved it but could not amplify it, or whether he never achieved it at all.

…When the organization undertook programs, however, it turned only to trusted members with limited tacit knowledge and presumed they could bring relevant talents to the task. The inadequacy of this “tribal” talent for some tasks caused diversion, waste, delay and a greater likelihood of failure.

Thus, for example, though Aum’s biological program conformed to the dominant understanding of the time by seeking to use bacteria as weapons, Endo was a virologist, not a microbiologist trained to work with bacteria. His skills with bacteria appear to have been more conceptual than hands-on. In this light, it is not surprising that the program repeatedly failed.

[cf. Barriers to Bioweapons. Good report overall on tacit knowledge & biowarfare. This is relevant to the discussion over LLM risks: the Aum Shinrikyo chemist could make a lot of progress by reading papers and figuring out his problems as he went, but the bacteriologist couldn’t figure out his issues for what seems like what had been a viable plan to weaponize & mass-produce anthrax but where lack of feedback led it to fail. Which does sound like something that a superhumanly-knowledgeable (but not necessarily that intelligent) LLM could help a lot with simply by pattern-matching and making lists of suggestions for things that are to the human ‘unknown unknowns’.]

See Also: