We Are Not Fully Aware of Everything China Can Do with Our Data - El Confidencial
This interview with Dr. Samantha Hoffman, Associate at Earendel Associates, was originally published in Spanish by El Confidencial and has been translated into English with the assistance of artificial intelligence. The translation has been prepared for accessibility and convenience; the original Spanish article remains the authoritative version and can be found here.
For Samantha Hoffman, an analyst at Earendel Associates, data is not merely a question of privacy. It is an instrument of power. Hoffman has spent years studying how the Chinese Communist Party turns technology, data and digital infrastructure into tools of security and influence, both within and beyond China.
Her work has helped change how the West approaches an increasingly uncomfortable question. Hoffman has testified before intelligence committees in the United States Congress, the UK Parliament and the European Parliament on what happens when the data we generate as we live, work and entertain ourselves acquires strategic value.
From video games and health data to critical infrastructure, Hoffman warns that apparently innocuous information can reveal vulnerabilities far deeper than we might imagine. El Confidencial interviewed her during a brief visit to Madrid, where she was sharing her expertise with several Spanish companies.
QUESTION. This week, it was reported that “malicious actors”, including China, have attempted to use Anthropic’s artificial intelligence to design and develop physical and biological weapons, or to support the collection of intelligence. How do you assess this emerging landscape, in which technologies such as AI have both civilian and military applications, and the accompanying sense that we are approaching some kind of “end of the world” scenario?
ANSWER. The unfortunate reality of emerging technologies is that they are all inherently dual-use. By that, I do not simply mean that everything can be used for good or ill. The same applications that were originally created for normal civilian purposes can potentially change at any moment.
They may not even need to change. They can continuously serve objectives connected to the concept of “national security”, the definition of which can be extremely broad, depending on the country concerned.
AI is a good example, but we need to think more broadly. Any technology that collects and processes data can potentially be used to access and exploit information in order to undermine an adversary’s interests.
One example would be a surveillance camera. A less obvious example would be smart-city technology that helps deliver public services. The problem is that these functions cannot simply be switched on or off. Data itself, which underpins the future of all these technologies, can be processed in many different ways. Its use ultimately depends on the intentions of the end user with access to it.
Similarly, many modern information technologies can be manipulated and used to contaminate the information environment. AI is undoubtedly one such technology.
Q. Are we being too paranoid, or not paranoid enough?
A. I would lean towards the latter. We are not paranoid enough.
Unfortunately, many of these technologies have already been deployed in numerous ways, making it extremely difficult to reverse course. Another part of the problem is that, as we continue trying to catch up with rapidly developing technology, many regulations have not been updated quickly enough.
Consider privacy laws. The average user’s ability to understand how their data is being used or processed is not sufficiently sophisticated to address these kinds of challenges.
Another issue is that, particularly in the West, we continue to think about security and defence, economic development and other relevant uses of technology as separate compartments. In China, they tend to overlap and merge in ways that we were not prepared to understand fully.
Societies are beginning to catch up. China is, of course, not the only source of concern. A number of private companies, including American companies, also collect and use data in ways that may undermine security. Nevertheless, I do not think we have yet fully understood what this means in terms of how we should respond.
Q. There appears to be a widespread attitude that says: “Yes, they are collecting my data, but so what? I do not have much to hide.”
A. It is difficult to imagine why you might be relevant as an individual, unless you are a government official, activist, member of the military or journalist, all of whom may feel that they have a target on their backs.
The point, however, is that data in aggregate can tell an adversary a great deal about many different things, far more than we might imagine.
Video games, gaming applications and future technologies such as virtual reality help reveal how individuals make decisions. When considered collectively, that data can be extremely useful in understanding how particular demographic groups make decisions.
There are also potentially more concerning uses of data, such as health data. In aggregate, health information can tell an adversary where particular weaknesses lie or what vulnerabilities exist within a country.
Although that may sound futuristic, or even somewhat fantastical, these potential applications of the mass collection, use and analysis of data are things that we should be considering now, before the risks materialise.
Q. You focus on the risks presented by China, but Palantir, an American company, has a £330 million contract with the UK health service. On one side, we have China, where the state exercises greater control over companies and their data. On the other, we have private American companies that also have relationships with their government and may presumably act in its interests. Europe is left in limbo, unable to place complete trust in either.
A. I think it is healthy not necessarily to trust any actor with access to large volumes of data.
Unfortunately, it is difficult to reverse the fact that enormous quantities of data are already circulating and being used for many different purposes. If you do not understand how your own data is being used to influence your decisions, that should be a cause for concern.
That said, some progress is being made in allowing individuals to opt out of particular uses of their data.
Q. But you have mentioned surveillance cameras and health data, as well as the possibility of governments using AI to cross-reference databases. In those circumstances, individuals cannot opt out.
A. Individuals deserve the right to understand how their data is being used and who has access to it. However, we do not have the appropriate systems in place, and I think we are a long way from having them.
Q. Will Europe’s strategy therefore be to regulate more, rather than beginning to make use of data for economic development, defence and other purposes? What should the European approach be? Opting out also means falling behind technologically, does it not?
A. Yes, it does mean falling behind. Another problem is that it creates a substantial burden for companies attempting to use data for more altruistic purposes. That is the balance we constantly need to strike.
This is where the discussion about whom you allow in and whom you block becomes important. Unfortunately, one of the few ways to deal with adversarial actors that have access to data is to try to prevent those channels from existing in the first place.
That requires a much clearer understanding of data supply chains. We need to understand when data is collected, exactly where it goes and who gains access to it at later stages, because the actor with access is not always the one you expect.
Healthcare provides a useful example. When you give your data to a doctor’s surgery, or allow it to be collected there, you do not necessarily imagine that Palantir might subsequently purchase it.
The same applies to genetic and ancestry testing. DNA data may ultimately be sold to third parties. There has been an example of a company selling such information to China. Having your genetic data end up in Beijing is probably not what someone has in mind when trying to discover whether they have Nordic or Viking ancestry.
Q. That data also remains stored, does it not? Even if it is not useful now, it may become useful in the future.
A. Yes. In fact, during the past two weeks, China enacted a revised version of its national defence mobilisation law, which will come into force on 1 October. One of its new provisions introduces a requirement to accumulate data.
Domestically, that data might be used to understand operational environments, environmental risks or other risks within an urban setting. In the context of national defence, some of it could support crisis-response planning, but it could also be used to gain a better understanding of vulnerabilities.
If China is involved in another country’s critical infrastructure, for example, all the resulting data may potentially be accessible to Chinese entities. It can then be processed and used to understand how the party-state might formulate its strategy. That is why data was included in the revised legislation.
Q. There is a perception that China collects large amounts of information and data primarily to exercise domestic control over its population, but that it has not applied this capability externally through disinformation or hybrid campaigns in the same way as Russia. Are there examples of China acting in this way abroad?
A. There is a misconception that China’s political-security objectives are exclusively domestic. In reality, the Chinese Communist Party also sees threats to its political security as originating outside the country.
Any alternative to the Party’s version of the truth is described as a threat to its political security, and that concept does not stop at China’s borders.
When it comes to data and its use, insufficient attention is paid to the effect of Chinese propaganda and information campaigns around the world. What China does successfully is identify important ways to insert itself into debates taking place within Western societies and use the language of those debates to advance its own interests.
Following Covid-19, for example, there was a wave of racist incidents directed against Chinese and Asian people more generally. The Party attaches itself to anti-racism campaigns in a way that is focused more on its own power than on the underlying issue, monopolising and polarising the debate.
Its approach is different from Russia’s, but we can see it happening and there is substantial evidence. Perhaps we are not always looking in the right places to understand it fully. It is not merely an information campaign on Twitter. It is a spectrum of coordinated activity guided by political objectives.
We are beginning to see that China is quite effective in influencing how we talk about certain subjects. The clearest example is the status of Taiwan.
Q. American AI companies have accused Chinese models of using their language models to build their own. Export controls and Western sanctions are intended to limit technology transfers, but they are clearly ineffective. Rather than slowing China’s technological development, is the West accelerating its pursuit of self-sufficiency?
A. Yes. One of the greatest perceived threats articulated by China is its dependence on foreign technologies within its own technological ecosystem.
The focus on self-sufficiency is not simply a high-level political objective. It also increases Chinese companies’ investment in research and development.
Although China has not succeeded in every area of technology, it has made rapid advances over the past decade, increasing its competitiveness and reducing its dependence on Western technologies. This has been driven domestically by the Chinese government.
Do sanctions and controls work? Unfortunately, not particularly well. There is considerable evidence of Chinese companies circumventing export controls through obfuscation and other techniques, including by gaining access to restricted semiconductor chips.
Q. Is there anything that can be done?
A. Part of what we need to do is develop a better understanding of supply chains and how Chinese companies operate through hundreds of subsidiaries under different names, making it difficult to identify with whom an organisation is actually interacting.
Unless we bring that under control, we cannot ensure that policies such as export controls are effective.
Q. Sanctions that do not work, as with Russia. Chinese state-owned companies have reportedly been sending dual-use materials to Russia’s military-industrial complex while classifying them as civilian goods in customs documentation.
A. Yes. This week, it was reported that a Chinese company was sending drones capable of military use while describing them in customs documentation as nylon-type materials.
Regardless of whether China and Russia merely have a close relationship or are direct allies, the war in Ukraine has shown that Beijing is willing to support malign actors. It has repeatedly demonstrated its willingness to help sustain Russia’s war effort in Ukraine.
Q. The war in Ukraine is probably an unparalleled source of military data. Almost every fatal drone strike is recorded by the drone carrying it out. Is there any known transfer of that data from Russia to China?
A. Whenever a crisis or event allows technology to be used on a large scale, it provides a real-time test that generates valuable feedback. That feedback helps China and other actors iterate and develop more effective solutions.
Even partial success or limited effectiveness in military technologies can be extremely damaging and have catastrophic consequences.
No such transfer has been reported. Nevertheless, one would expect any technology supplied by China to Russia probably to include provisions enabling data to be collected and returned to improve the product and the quality of the service.
That is normally where such provisions are embedded, in documents such as the privacy policies of Huawei or TikTok. Information may be transferred to the company and its subsidiaries to be processed for ordinary business purposes. Under these privacy policies, however, those ordinary purposes often encompass China’s broadly applied conception of national security.
The same principle would apply to technologies exported for military use.
Q. Spain has been at the centre of controversy with the United States over alleged contracts with Huawei that provided access to data. What would your advice be?
A. Several steps could be taken to improve the situation.
One would be to invest in understanding the networks and vulnerabilities created through the relationships that critical companies maintain with Chinese businesses. The country must then decide what level of security vulnerability it is, or is not, prepared to tolerate when engaging with those actors.
Spain must decide for itself what it is willing to accept, based on very clear definitions of how an adversary understands access to information or critical infrastructure.
If we understand that more clearly, decisions should become considerably more risk-averse. However, there is also a great deal to unravel in relation to decisions that have already been made.