Unlike Pegasus, Zhenhua lacks prowess to spy; China's claims on data harvesting a smokescreen, exposes its weaknesses
Zhenhua isn’t important for what it does, but what it tells us about China’s weaknesses and insecurities.
Ensconced in their dour winter jackets, the red of carnations clasped in calloused hands was the only colour interrupting the grey ranks of the faithful, lined up on the second Sunday of January, 1988, to march in memory of the Berlin communists by Nazis. Vera Wollenberger waited by the roadside, hoping to hold up a placard of protest with the seditious words of the German Democratic Republic’s constitution emblazoned on it: “Every citizen has the right to express his opinion freely and openly.”
The police were waiting, too: For more than a month, Wollenberger was held in Berlin’s Hohenschoenhausen Prison.
Laid end-to-end, the files of the German Democratic Republic’s Ministerium für Staatssicherheit—popularly, the Stasi—run for 111 kilometres, documenting in intimate detail the lives of some 5.6 million people. In 1991, the Stasi had 91,015 staff, and at least 173,081 paid informants —- more than two for every 100 citizens.
The man who fed Wollenberger’s own file was her much-loved husband, Knud Wollenberger. In secret, the eminent mathematician and poet was a Stasi agent who, driven by his passionate anti-Fascist convictions, had agreed to report on their dinner conversations and their pillow-talk.
Less than a year after the GDR exiled Wollenberger, though, the country itself collapsed. “When a country is its own worst enemy,” the author Craig Whitney has noted, “having the best spy service can’t help”.
EVIDENCE that the Zhenhua Data Information Technology has been trawling the internet for data on more than 10,000 Indian nationals, revealed by The Indian Express and an international coalition of journalists on Monday, has underlined growing fears about China’s information operations. The company, leaked files show, collected data on more than two million prominent individuals worldwide. Zhenhua’s software then applied Artificial Intelligence-driven tools to this data, looking for links and patterns useful to understanding their behaviour.
Zenhua’s operations have been widely described using phrases like Fifth Generation Warfare — wars of ideas, rather than bullets — but there’s good reason to suspect it is actually been peddling snake-oil to China’s intelligence service, the Ministry for State Security.
For one, the raw data collected by Zhenhua is of little real value. In essence, the company has trawled the internet for information that can be found with nothing more sinister than Google. Pegasus, the spyware whose misuse by law-enforcement and intelligence services was outed earlier this year, penetrated encrypted communications channels. There’s no suggestion that Zhenhua has those capabilities.
Zhenhua likely told its clients that it can create replicate the country’s massive domestic databases at a global level. AI, its sales-pitch likely went, could then generate analytical output superior to human analysts or social scientists. Instead of depending on overworked analysts at the MSS, the job of thinking could be left to machines.
There are, however, questions about the utility of Zhenhua’s AI-driven analytical output. Experts like Princeton’s Arvind Narayanan have pointed out that while AI is excellent at certain kinds of tasks, it is poor at predicting social outcomes or complex individual behaviours. Even in relatively straightforward tasks, like medical diagnostics, there is growing scepticism over the claims made by AI companies.
For decades now, experts on Chinese intelligence like Peter Mattis have noted, its spies have struggled to address “the blind spots created by the country’s domestic-based intelligence posture”. The problem has been aggravated, Mattis noted in 2016 testimony to the United States Congress, by the fact that China’s “clandestine tradecraft probably does not rate among the world’s most sophisticated”.
Like China, global intelligence services have become increasingly invested in technology hoping to address human-capability deficits. From 1947, the five English-speaking powers—the United States, United Kingdom, Australia, Canada and New Zealand—signed the so-called “Five Eyes” treaty, allowing for the sharing of intelligence. Ever since the 1970s, Five Eyes satellites fed a system known as Echelon, which today sucks up virtually all electronic communication from around the planet—all ending up in the United States National Security Agency’s computer-farms.
Edward Snowden, the NSA whistleblower, revealed that between 8 February and 8 March, 2013, alone, a single NSA tool called Boundless Informant collected about 124.8 billion telephone data items and 97.1 billion computer data items throughout the world.
The single largest employer of pure mathematicians in the world—hiring more PhDs than India produces in a decade—the NSA has the capability to break large-prime number encryption, the bedrock of online privacy. Like India, China is believed to be seeking similar capabilities, though so far without significant success.
Fashionable clichés about data being the new oil, though, elide over an unpleasant truth: mere information isn’t actual knowledge. The NSA’s penetration of Al Qaeda communications didn’t stop 9/11. Even though the United Kingdom’s surveillance of Lashkar-e-Taiba computer networks generated warnings, India was unable to preempt 26/11. Total intelligence domination didn’t win the United States its wars in Afghanistan or Iraq. Intelligence services drowned in information, failing to distinguish gold from the dross.
In 1974, the great GDR spymaster Markus Wolf wrote in his diary: “almost all the reams of paper produced by NATO and stamped with the codes ‘cosmic’ or ‘top secret’ are, when you get right down to it, not even worth using as toilet paper”.
Wolf remained deeply conscious of the limitations of technology-driven espionage: “Technology can only establish the situation of the moment,” he argued, “[but] secret plans, options and other considerations will remain concealed even from the most sophisticated satellite”.
Potentially war-winning intelligence, critically, has often come not from technology, but spies like Richard Sorge, who provided precision warnings on Nazi Germany’s intention of attacking the Soviet Union.
Large volumes of information, history teaches us, often works to bog down an assessment—and thus undermine the raison d’etre of intelligence. Emperor Sultan Abdulhamid II ran perhaps the largest intelligence service in history, recruiting clerics, clowns, illusionists, dervishes, and even a beggar from India to monitor his creaking realm. “Everybody began to report on each other,” Ekrem Ekinci has observed of the collapse of the Ottoman surveillance state. “Absurd rumours and even slander began to be reported”.
“In time”, Ekinci noted, “the value of these reports diminished”. Thousands of unopened missives were discovered when Turkish revolutionaries overthrew Abdulhamid II in 1908—a move the sultan’s intelligence network proved unable to preempt.
Wollenberger’s helps understand why large-scale surveillance can actually counterproductive. Andreas Lichter, Max Löffler and Sebastian Siegloch found the “density of informers undermined trust and led to a withdrawal from society”. “In particular, more intense surveillance caused lower trust in strangers, stronger negative reciprocity, fewer close friends, lower sociability, and reduced societal engagement”.
The scholars noted the “negative and persistent effects of government surveillance on various measures of economic performance, such as individual labor income, county-level self-employment, unemployment and population size”.
For a country like China, working to make the transition from low-grade manufacturing to technological creativity, intrusive surveillance can thus be an impediment to progress. China’s mass surveillance in Xinjiang, for example, hasn’t demonstrated stellar success: mass data has helped incarcerate hundreds of thousands of people in concentration camps, without actually addressing the problem.
Even two thousand years ago, the costs of the surveillance state were evident. “The breed of ears and provocateurs make tyrants, who are obliged to know everything, most detested”, the classical scholar Plutarch wrote. The root problem, he noted, was the insecurity of rulers, not the secrets of their subjects.
Like Plutarch’s tyrants, the Chinese state believes its power rests on seeing and hearing all. The babble its spies are gathering online, though, most likely makes it impossible to actually listen to what is of importance. Zhenhua isn’t important for what it does, but what it tells us about China’s weaknesses and insecurities.