A long read but worthwhile one. Introduces the concept of cyops, or cyber-psychological operations which may be AI assisted.
While, until recently, cyber operations have constituted a specific subset of defense and security concerns, the synergization of cyberspace and artificial intelligence (AI), which are driving the Fourth Industrial Revolution, has raised the threat level of cyber operations, making them a...
www.intechopen.com
Excerpt:
"As stated previously, in the present chapter,
cyops are a major part of hybrid operations, and
cyber psychological tactics involved in
cyops typically include [
3]:
- Propaganda (in particular, dispersed online through social media)
- Fake contents (in particular, fake news)
- Online dispersal of sensitive data (leaks)
Each of these tactics takes advantage of network power, AI power, and cooperation power. There are three drivers that have amplified the effectiveness of the above tactics:
- The increased dispersal of connected devices, including smartphones and tablets that allow an easy and frequent access to the Internet
- Search engines and online services that adapt to each user’s interaction pattern
- The growing use of social media over traditional media
Added to this infrastructural accessibility to these devices is the high frequency use of these devices and sometimes addictive component associated with this use, an addictive component usually linked to social networks.
A specific pattern of usage favors the dispersal of sensitive data, news, and general contents in social media: the fact that the online reading of social media contents usually does not involve a high level of reflection but rather engages the users in a way that is meant to be appealing and to be shared quickly with as most people as possible, users seldom read or reflect deeply on the contents that they are sharing, usually skimming through them and sharing the most appealing ones.
This is a pattern that is particularly useful for dispersal of contents that are presented in the form of scandals, sensitive information that was not known, conspiracies’ denouncements, and so on. This point leaves a marker in data on fake content dispersal as shown in a study on the differential diffusion of verified (true) and false rumors on Twitter from 2006 to 2017, published in [
12]. In the study,
politics and
urban legends stand out as the two categories with the highest frequency in rumor cascades.
The study concluded that rumors about politics, urban legends, and science spread to the most people, while politics and urban legends exhibited more intense viral patterns [
12].
The study found a significant difference in the spread of fake contents vis-à-vis true contents, namely, true contents are
rarely diffused to more than 1000 people, while the top 1% of fake rumor cascades are
routinely diffused between 1000 and 100,000 people [
12]. The authors’ results showed that fake contents reached more people at every depth of a cascade, which the authors defined as
instances of a rumor spreading pattern that exhibit an unbroken retweet chain with a common, singular origin.
The result that fake contents
reached more people at every depth of a cascade means that more people retweeted fake contents than true ones, a spread that was amplified by a viral dynamics. The authors found that fake contents did not just spread through broadcast dynamics but, instead, through peer-to-peer diffusion with viral branching.
Another relevant point, for hybrid operations, was that fake political contents traveled deeper and are more broadly reaching more people and exhibiting a stronger viral pattern than any other categories and diffusing deeper more quickly. This dynamics is not however due to users who spread fake contents having a greater number of followers; the study found exactly the opposite with a high statistical significance. In inferential terms, users who spread fake contents tend to have fewer followers, to follow fewer people, to be less active on Twitter, are verified less often, and have been on Twitter for less time. However, fake contents were 70% more likely to be retweeted than true contents with a
p-value of 0.0 in Wald chi-square test.
The fact that user connectedness and network structure did not seem to play a relevant role in
fake content dispersal made the authors seek other explanations for the differences in
fake content versus
true content dispersal. The authors reported that fake contents usually inspired greater number of replies exhibiting surprise and disgust. The authors’ hypothesis is that novelty may be a key factor in
false rumor dispersal.
However, there is a relevant point to take into account when looking at the study’s results, which can be expressed by the following extreme example: an account with no followers and not following anyone can still get a high number of retweets and exposure on a content if it uses
hashtags on hot topics and builds its tweet in a specific way that increases the probability of it being retweeted.
Moving beyond this specific study and considering social networks in general, working with the conceptual basis of strategic studies, we are led to introduce the concept of
tactical accounts, defined as accounts that are created for tactical purposes in the support of a
cyop strategy; these accounts can be managed by a single individual or staffs, and its operations can involve the use of
bots that automatically generate contents with certain specifications, mostly aimed at making the contents viral in the spread.
The viral content design along with multiple accounts operated by
bots are major tools for a
tactical account system manager, that is, any operative can use multiple tactical accounts simultaneously to create a fake content dispersal so that it can gain momentum and become viral.
In general, fake contents can spread on hot topics by the use of
hashtags or other means of dispersal, which diminishes the connectivity need for any single
tactical account’s effective impact. Furthermore, from a
cyops’ standpoint, it is easier to
fly under the radar by managing multiple newly created fake accounts that can even be managed by a single agent, who may then use these accounts to disperse fake contents incorporating
hashtags on political issues and composing the messages so that they have an appealing emotive content, making it more likely for people to select them.
...
Empowered by
cyop bots, a small number of individuals, or even one individual, can create a game that may go on independently of them; the game can persist as a dynamics that continues to be played in the platform, which functions as a replicator for the deviant and predatory behavioral patterns needed for the
terror game to go on. Having been played once, the dynamics that characterize the game can always come back; in this sense, the platform works as a way for the digital continuation of the terror game.
This is very different from the case of a terrorist network that has a hierarchical structure and that has cells and individuals that play different roles within an organization.
A
terror game is just a set of behavioral patterns, with algorithmic components, that can be replicated like a form of social virus which goes on as long as there are players. There is no stable hierarchy and no cells and no individuals that can be targeted which may harm the game, because the game has a virtual fluid existence that can be perpetuated as a dynamics to be retrieved any time, any place.
The
terror game is characteristic of a side of platforms, especially social networking platforms that make them highly weaponizable, namely, platforms are means for the exercise of biopower in the sense of Foucault [
21], a point that is convergent with the issues addressed in [
22].
Platforms can function as means for the exercise of control, reward, and punishment and of manipulation of its users’ desires, fears, and sources of inclusion and exclusion, integration and segregation, connection and isolation, and friendship and bullying.
By increasingly sharing one’s life in platforms and by using integrated systems, in particular IoT devices, the new stage of the Internet revolution is such that any heavy user of these systems can be
datafied, profiled, and manipulated by hacked devices (including hacked AI systems) and manipulated by predators that use fake accounts and their victims’ profiles to launch directed
cyops that can, in the end, as was the case with the
Blue Whale Challenge, lead to a person’s death.
According to data, reported in [
20], Instagram ranks higher in posts than the Russian VK social network (which was where the game spread initially) and Twitter. On Twitter, the large majority number of posts related to the
Blue Whale Challenge was identified by the authors as coming from
smartphones with the Android OS, which shows how mobile devices are useful in feeding
terror gamification operations.
Another pattern revealed in these authors’ research is a key common factor in online
cyop campaigns. In particular, many accounts talking about the
Blue Whale Challenge were new accounts with not many followers; this shows again the possible use of
tactical accounts. This is a basic necessary tactical choice for predators operating online, who will want to hide their identity; furthermore, in order to gain online traction on a
cyop, the use of multiple
tactical accounts is a necessary step. Thus, just as in
state agents,
non-state agents, including
cyber-enabled serial killers, may tend to use multiple
tactical accounts in online platforms when addressing their targets.
The use of challenges like the
Blue Whale Challenge and the
Momo Challenge directly targets a large amount of victims and constitutes a security and law enforcement problem [
23].
Returning to
cyops, whatever their profile, these are currently about using cyberspace and ML for
hacking people’s behaviors. In this sense, while a
cyop against a given country may take advantage of resilience, authority, or legitimacy vulnerabilities, the increasing use of the platform-based technologies, managed by ML algorithms, where each user’s data is exposed and available for exploitation, leads to another level of vulnerability which is the ability to use citizens’ own data and behavioral patterns against them or to manipulate citizens into patterns of behavior that interest a given state or non-state agent."
m.washingtontimes.com
