In November 2007, two teams of intruders broke into the South Africa nuclear research station, Pelindaba, located west of Pretoria. One team cut through the high-voltage fence at a point where video surveillance cameras did not reach and headed to the installation’s electrical box, where a magnetic anti-tampering mechanism was circumvented and the power to a fence and its alarms were deactivated. By chance, the first team was detected by a visiting off-duty fireman, who was assaulted and shot, after which the team absconded. The second team, meanwhile, had not got far into the installation before it was apparently warned and retreated.
Pelindaba is a nuclear research centre established in 1965 and run by the South African Nuclear Energy Corporation (Necsa), a state-owned public company. The official purpose of Necsa was to the carry out research and development of nuclear energy and related technologies, to process and store nuclear material and to coordinate with other organisations concerned with these matters.
Pelindaba’s role was ostensibly to develop nuclear power for peaceful purposes, but by the mid-1970s it was reportedly processing weapons-grade uranium. The then-government denied making weapons of mass destruction, but by 1989 it was revealed that it had built six weapons, which were dismantled. In 1993, then-president FW de Klerk publicly admitted that the country had instituted a nuclear weapons programme.
As a consequence of that apartheid-era programme, South Africa possesses a stock of highly enriched uranium (HEU). The stock is considered a strategic national asset, and is stored in a vault at Pelindaba. The exact amount remains classified, but experts estimate an amount of between 450 and 760 kilogrammes (Civilian HEU, 2019). Some of this HEU has been used to fuel Pelindaba’s research reactor; small research reactors typically run on between 2kg and 5kg.
HEU can be used to make crude nuclear weapons, and is “more difficult to detect by technical means … and much easier to divert, smuggle and hide” than plutonium, making it the material “most wanted by terrorists”, according to the International Nuclear Information System. Experts estimate the amount of HEU needed for a terrorist bomb at between 25 and 50kg, depending on the extent of enrichment. Taking these estimates together, the amount of HEU held at Pelindaba would be adequate for between 8 and 30 bombs (Bunn & Braun, 2003).
Given these facts, the raid on Pelindaba raised an international outcry, though somewhat underreported at the time. It was only when a swathe of secret spy communications was leaked in 2015 — the so-called Spy Cables scandal — that some of the serious implications of the Pelindaba raid were revealed in more detail. In outline, the leak revealed that South Africa’s intelligence capability at the time was severely lacking at best.
The Pelindaba break-in: accounts of what happened
SA intelligence failures on the raid
According to a CBS News documentary based on the Spy Cabes leak, a former director of the US’ CIA had said that the aim of the raid was likely terrorism. This was hotly denied by the South African government, which said that the break-in had been a case of “ordinary theft”. A spokesperson said that the risk interpretation was a “campaign” to “undermine” the government by turning a burglary into a “major risk”, adding that the concern was based on “conspiracy theories” and “innuendo”. Yet the facts of the case pointed to more than conspiracy.
An independent investigation commissioned by Necsa, only revealed during the leak, concluded that the raid relied on insider information, was carefully planned and had probably targeted the nuclear material. The investigation used cellphone records, interviews and polygraph tests to identify two South Africans who might have been accomplices, but the information was apparently not taken up by the official investigation.
National and police intelligence services appear to have conducted a pro forma investigation into the break-in at best. Three suspects were arrested in connection with the raid but nothing further was heard about them. A Malawian who was detected using a SIM card from a cellphone that had been stolen during the raid was deported.
Then it emerged that at the time, South African spies had told the Americans that the Chinese were suspected of involvement. This view was at odds with the “burglary” theory, since it was based on the idea that the Chinese wanted information about the South African pebble bed modular reactor technology. That alone would have made the raid more than an “ordinary burglary”. CBS said that other investigations had left the South African government feeling “sensitive about [the country’s] nuclear security”.
Necsa suspended or fired some Pelindaba security personnel and improved its security, based in part on the secret report. An International Atomic Energy Agency team inspected the installation and concluded that no sensitive nuclear areas were breached during the break-in. Though probably true, this effectively discounted the fact that the raiders had already accessed a laptop in the centre and were apparently set on gaining access to other areas before they were interrupted. Years later, the case was being cited in a Stanford University course as a paradigmatic example of nuclear security failure that posed “grave security concerns for the world”.
As it turns out, the South African government had good reason at the time to be “sensitive”, or plain shirty, about the revelations into the Spy Cables scandal. Suspicions that the South African intelligence function was being subverted to party-political purposes had been surfacing for some time prior. In 1998, then-South African National Defence Force chief General George Meiring claimed that a “left-wing plot” run by ANC veterans was working against his command of the military. A commission of inquiry dismissed the claim, but its argument that the claim was based on a disinformation campaign by former apartheid securocrats itself smacked of disinformation. In October 2005, South African businessman and ANC veteran Saki Macozoma claimed that he and his family had been surveilled by the National Intelligence Agency (NIA) on suspicions that he was part of a “plot” against then-president Thabo Mbeki.
By then it was already evident that the intelligence services were becoming severely politicised. A 2006 report by the Office of the Inspector General of Intelligence concluded that claims of “counter-revolutionary threats” were contributing to internal ANC party disputes. The pattern of politicisation appears to have intensified with Jacob Zuma’s presidency, from 2008. Ronnie Kasrils, a veteran of the ANC underground struggle and intelligence minister from 2004 to 2008, resigned because of the “shenanigans” he was observing in the intelligence services, which apparently originated in the conflict between Mbeki and Zuma. Another former spy chief, Barry Gilder, said that the rivalry had begun in the early 2000s, when Zuma was vice-president under Mbeki and divided an “already fractured spy network”.
The increasing repurposing of the intelligence services for party-political purposes had a direct impact on South Africa’s fledgling democracy when it was revealed that intelligence officials had secretly recorded conversations between the then-head of police crime intelligence and the then-head of the National Prosecuting Authority (NPA) about Zuma’s alleged corrupt involvement in the arms deal of the late 1990s. Zuma, indeed, has used the fact that these conversations were illegally recorded as an argument in his decades-long so-called Stalingrad approach to staving off investigations into his involvement.
Indications that the intelligence services were being used in internal power struggles continued to emerge. In 2013, claims emerged of a supposed “rogue unit” comprising former and serving officials of the NIA that was alleged to have conducted intelligence investigations within the South African Revenue Service (Sars). In August 2021, the former minister of Human Settlements, Water and Sanitation, Lindiwe Sisulu, registered a complaint against the State Security Agency (SSA), claiming that the communications of departmental officials in the Eastern Cape had been intercepted and that the officials had been “harassed” (Cave & Jurgens, 2021, 73).
Investigations into SA intelligence
Suspicions concerning the intelligence services were investigated by at least three high-level teams over the long period under review. These include the 2008 ministerial review commission, the 2018 high-level review panel on the SSA, and the Zondo commission, which completed its hearings in August 2021.
The 2008 ministerial review commission report noted that critical areas of the intelligence legislation required review. It recommended changes to legislation to ensure that the intelligence minister’s access to, and powers over, intelligence were regulated. It also proposed regulating the access of other government departments to intelligence, regulating the access of the president to intelligence, and providing for measures to discipline or dismiss the heads of intelligence structures. Remarkably, the latter power, in particular, had not been thought of, even though it was obvious that such roles included access to highly sensitive information that could be misused and in which in case requires formal protection in terms of law.
The 2018 high-level panel report concluded that the civilian intelligence function had become “a private resource to serve the political and personal interests of particular individuals”. It noted that “a special operations intelligence unit, set up under Zuma’s watch, undertook operations that were ‘clearly unconstitutional and illegal’”. The intelligence function had been “seriously politicised for internal factional purposes for a decade or more, “in an almost complete disregard for the Constitution, policy, legislation and other prescripts”. Moreover, it policy as regards the intelligence services had shifted from a focus on national security to a focus on state security.
In 2021, whistleblowers testifying to the Zondo commission revealed that for nearly ten years, between 2009 and 2018, ministers in the cabinet had and regularly run operations, handling sources, tasking them, debriefing them and generating reports “possibly to advise themselves, the president or their colleagues”. This would have been in defiance of constitutional and legislative measures designed to ensure a separation of the executive from key government operations. As part of this, it was revealed that some R9-billion in fixed assets and R125-million in cash could not be accounted for in the 2017-18 financial year alone. Parliamentary oversight of the intelligence function had been “weak or non-existent”.
If anything is clear in this murky tale, it is that no one involved felt any consequences. According to testimony at the Zondo commission, the SSA failed to submit an annual report for three consecutive years. Cabinet members abused the intelligence services for party-political purposes and operatives treated the intelligence funding as what journalist Marianne Thamm called a “giant ATM with no pin code”. Yet it appears that no one has been disciplined, suspended or fired for these crimes.
The problems with intelligence were so intense and so far-going that they were said to be at the root of the state’s complete lack of preparedness for the July riots of 2021, which devastated the country. “Reasons include lack of vision, political interference and state capture, bad management, poor coordination, weak policy and insufficient accountability such as through parliament… It is clear that South Africa has an intelligence service in name only”.
The above is a summary of the failures of the South African intelligence services in relation to the July riots. But it appears that serving cabinet ministers have ignored constitutional and legal prescripts over a long period. The summary therefore usefully reflects the political and institutional failures that have seen a serious subversion by members of the ruling party of the aims and protocols that are supposed to cover the governance of the intelligence services.
It is worth observing that the ongoing subversion of the country’s intelligence services over a long period has been enabled not merely by political or institutional “shenanigans” but also by what appears to be a concerted tactic of an undermining of basic truth and meaning.
Under Zuma, the former National Intelligence Agency became the State Security Agency, for example. This semantic reframing of the role of the intelligence services effectively allowed members of the ruling party and their minions to identify their own interests with those of the country. And they were able to do so in the plain light of day, because no one thought to examine the semantic subversion at work.
In his political satire, 1984, George Orwell famously developed a notion of an official language for the authoritarian state, called Newspeak. The purpose of Newspeak was “to diminish the range of thought … by cutting the choice of words down to a minimum”. Newspeak, that is, was a constructed language that reduced the range of meanings available to a bare minimum, thus controlling thought. In South Africa, the language used in the course of official abuses of the intelligence function in South Africa, and indeed more widely in the ruling party’s pursuit of its endlessly redefined agendas, suggests a variation of this: NoSpeak.
A good example is another scandal revealed by the 2015 Spy Cables leak. It emerged that South African intelligence services had agreed a deal with Zimbabwe to spy on “rogue NGOs” aimed at “subverting constitutional order”. As regards South Africa, various official investigations have established that those charged with the management of the intelligence services were themselves subverting the constitutional order over a long period. It is also universally clear (outside of Zanu-PF circles) that Zimbabwe’s continuing oppression of the country’s people is based on an outrageous perversion of constitutional principles. If this subversion of meaning were not so serious, it would be laughable.
As opposed to Newspeak, which restricts meaning to an authoritarian minimum, the aim of NoSpeak is to say nothing while appearing to say something.
This tactic appears to assert something definite and accountable while in fact leaving the supposed communicator room to do what they like. This is the technique of semantic subversion, which has recently been defined as a rhetorical manoeuvre that serves “to establish new and powerful belief systems beyond rational and democratic control”.
In fact, this definition probably doesn’t go far enough in understanding the phenomenon as it is practised in real-life contexts such as South Africa’s intelligence world. For it would appear that South African NoSpeak can better be understood as an applied semantic subversion, the aim of which is not to establish a new system of belief, but to undermine all systems of belief.
NoSpeak, here briefly defined as semantic subversion, involves at least three approaches to undermining meaning. One is Silent NoSpeak, which is practised when ministers and their officials ignore legally mandated requests for information in the form of Promotion of Access to Information Act requests.
A second form might be described as Lying NoSpeak, which uses language to appear to say one thing while knowing that the opposite is true, for example, appearing to support principles of governance while in fact subverting them. This might be said to be little more than lying, but the broader context of its use in undermining meaning needs to be taken into account.
This third form of NoSpeak is the most serious: Noisy NoSpeak. This occurs when ministers or government officials undermine the meaning of commonly accepted terms to create ambiguities about the meaning of the terms that prevents an understanding of what is really going on.
Perhaps the most trenchant example of Noisy NoSpeak is the concept of “transformation” as it has been applied to the intelligence services. After the demise of apartheid it was clear that the new government would face serious challenges in creating a new intelligence service that was suitable to a constitutional democracy. According to the Intelligence White Paper of 1995, the intelligence services needs to be “reshaped and transformed” to serve a democratic purpose. In its defence of its proposed Protection of State Information Bill, meanwhile, the ruling party argued that it had “followed a broader approach to national security”. Yet something like the opposite was true: as the 2018 high-level pPanel noted, the intelligence services were subverted to a narrower purpose, that of serving party-political and factional ends.
This is a form of Noisy NoSpeak, and not merely Lying NoSpeak, because the terminology relies on an understanding of “transformation” that is essentially undefined. In a wider context, President Cyril Ramaphosa appears to confirm that this technique of introducing fundamental ambiguity of meaning has been common within his party more generally. The term “radical economic transformation”, he said, “has often been misused, misrepresented or misunderstood”. A commentator on the Spy Cables scandal argued that “it’s become practically impossible to work out the competing factions that exist inside the SSA”. One insider to internal ANC processes told me that this mirrors the atmosphere of many meetings, where no one knows what the latest buzzword means, or which faction it represents, or indeed what the factions represent. In the inherently ambiguous world of intelligence, this effect seems to have been amplified to deafening levels.
Need for ethical governance
In an article published in the first issue of The Africa Governance Papers, Dr Lincoln Cave and I show that the formal measures embodied in the Constitution and relevant legislation “conform to international governance practice … [and] also align with the constitutional requirement that national security organisations are accountable to government, parliament, and ultimately, the public”. As we also argue, the extensive abuses of the functions of civilian intelligence entities … indicate that these formal arrangements have lacked effective governance bodies and oversight mechanisms”. We propose some practical measures to deal with these failures, including a clear model of ethical stewardship of intelligence functions and the need to integrate it into intelligence governance.
However, correcting on the wild abuses of the intelligence services over the past 15 years and preventing their further abuse will also require a deeper and more nuanced understanding of the processes of communication that surround them, and in particular, the government’s instinctive resort to semantic subversion to defend, obfuscate or enable those abuses.
The Pelindaba raid was a serious case in point. Whatever else is clear, it is that it is very unlikely that the that sophisticated incursion into a nuclear facility was a mere case of “burglary”. Even if the aim was not to gain access to dangerous nuclear materials, the raid showed that this might have been fairly easily been possible.
Finally, it is worth noting that the pebble bed modular nuclear technology that was being developed in the nuclear facility was also at risk. At the time, nuclear experts estimated the chances of success with the technology were as very low. The South African project had experienced problems for years, which, according to one report, centred on the challenge of containing the heat of the reactor and that the pebble bed modular nuclear technology might prove to be “fundamentally flawed” (Thomas, 2009). Yet only late last year, the Chinese announced that they had successfully developed and commissioned a pebble bed reactor. The project, which began in 2012, according to official Chinese sources, was now working well, and adaptable “to small and medium-sized power grids” — just the kind suitable to South Africa.
Who knows if the country lost an innovative technology through the internal failure and corruption of our intelligence services? It is not clear if the intelligence services themselves know. And if they do, it is likely that NoSpeak has effectively prevented them from realising it. In NoSpeak, it’s not strictly possible to know what’s true, since no one really knows what’s being referred to. And equally, it’s not possible to know what’s false, for the same reason.