The Use of Social Media Bots and Automated (AI Based) Text Generators: Key Technologies in Winning the Propaganda War Against Islamic State/Daesh?
This essay in the TRENDS Terrorism Futures series will discuss Islamic State(IS)/Daesh’s use of social media along with recent trends in commercial bot and automated text systems. It will then highlight the relationship of IS/Daesh to computer algorithms and similar bot-like applications in social media along with automated text generators. It will conclude with the implications these technology trends may likely have on IS/Daesh and on the international community response being directed against it. Ultimately, in such an increasingly computer science and artificial intelligence dominated conflict, IS/Daesh may find that replicating its early social media successes may now be extremely difficult.
IS/Daesh and Social Media
IS/Daesh has proven itself to be highly adept at using social media to promote its extremist ideology and expand into a global terrorism network that has displaced Al Qaida as the premier Sunni radical Islamist organization. Such media fulfills the need to widely communicate group activities, help radicalize dispersed communities of interest, raise funds online, and to ensure that a fresh supply of foot-soldiers replace those killed in airstrikes and ground offensives in Iraq and Syria. Social media is also critically important to IS because it can be used for psychological and propaganda purposes—including the terrorizing of besieged towns—and can help to promote IS international branding and market share as the rightly guided stewards of reestablishing the Caliphate.
Recent reporting on IS/Daesh’s use of Twitter suggests that in autumn 2014 this terrorist organization peaked with at least 45,000 accounts (an aggregate of both newly created and suspended accounts). The group had earlier developed the Arabic language ‘The Dawn of Golden Tidings’ (i.e. Dawn) app for Twitter for news group purposes and had also gained a mastery of hashtag use for tweet campaigns to obtain top trending tags which are then retweeted by the @ActiveHashtags stream.
IS/Daesh is an adaptive network and has seen almost 19,000 actual and suspected member Twitter accounts—mostly in English—closed as well as the shutdown of the Dawn app. As a result, since early 2015 it has shifted from English to Arabic as its dominant form of social media communication. Of course, the continued posting of pictures and audio-video imagery is still taking place—though now predominately in Arabic. Additionally, due to the recognition of ongoing IS/Daesh social media analysis by the international community, that terrorist organization has now placed malware triggers in its blogs. When Arabic text is translated into English and also Japanese via search engines such as Google and Bing, malware is released which infects the computer systems of unsuspecting researchers. Additional types of related IS/Daesh malware countermeasures—such as keyboard loggers and remote access software—also exist based on other triggering protocols.
Social Media Bots and Automated Text Generators
As of August 2014, Twitter was said to have over 270 million users who are active and engage in tweets. Of these users, roughly 23 million are bots and apps utilized for humor, commercial spamming (e.g. advertising and promotional links), and malware and virus insertion purposes. Bot creation is thus very prevalent with some websites—Swenzy and Fiverr, for instance—selling ‘likes,’ ‘comments,’ and ‘followers’ to social media accounts. Other forms of social media such as Facebook have similar issues with 140 to 170 million fake—that is bot derived—profiles and users estimated out of roughly 1.49 billion accounts (i.e. accounts logged into in the last 30 days).
Beyond these relatively basic bots and the bot nets that may, control them another level of computer science sophistication is beginning to have an influence on global social media. Automated text generators have developed within the last decade and are being spearheaded by a handful of high-tech companies. Two approaches are being undertaken for commercial purposes and are focused on exploiting structured data for mass news reporting and for book creation. In the news reporting realm, Narrative Sciences and Automated Insights dominate and have created AI systems that learn and benefit from ‘meta-writer’ templates (narrative shells and rules provided by subject matter experts such as journalists) which allows for the news text produced to be read by humans without them realizing that it is machine generated. In 2014, over a billion yearly sports updates of little league games and women’s baseball, weather forecasts, corporate earnings reports, monthly franchise reports, and similar material are being produced in this manner with the Associated Press, Forbes.com, and other news agencies now utilizing such text generation services.
The book publishing utilization of these systems has been pioneered by Icon Group International for the mass consumer market and its application for internal corporate reports (like the monthly franchise reports noted earlier) and even the licensing of such proprietary AI has been vested with its subsidiary EdgeMaven Media. Hundreds of thousands of formulaic books and other material have been created via this AI system which are both original and non-plagiaristic products but not necessarily creative in their approach. Essentially, the computer algorithms utilized scrape the internet of the specific topical content input via the parameters provided and then, using relational metrics and other heuristics and analytics, can quickly and cheaply create a book length product such as The 2007-2012 World Outlook for Wood Toilet Seats. Tables, figures, indexes, and imagery such as photographs can also be added to such products. While the book will likely sell very few copies at a price of $795.00, it only requires one sale to make a massive profit based on the fact it cost less than a dollar to generate. Repeat the process within other specialized book niches hundreds of thousands—what in the future will become millions—of times and the commercial value of this approach becomes readily apparent. Competing systems such as Nimble Book’s PageKicker system also exist as well as the experimental European What-If Machine (WHIM) project.
IS/Daesh Use and Potentials
As seen in the preceding section, from a terrorist perspective bots and auto text generators can offer new social media influencing potentials. Focusing on bots first, our understanding of actual IS/Daesh use of social media bots, however, is very much more opaque than other social media sectors. A best estimate is simply that thousands of Twitter bots are being utilized. To a lesser extent, Al Qaida in the Arabian Peninsula (AQAP) and Al Nusra Front in Syria (an Al Qaida affiliate) are also employing such bots but have proven far less adept at their use. Per the ISIS Twitter Census report produced by the Brookings Institution in March 2015, the IS/Daesh bot use strategy emerged after the demise of Dawn in June 2014 and is as follows:
In the wake of that setback, ISIS supporters have responded by creating a large number of bots in small clusters, with each cluster using a different service to post tweets of the propaganda and hashtags it wishes to promote. If one “family” of bots is suspended, there are still many others that will continue to tweet. Thousands of such accounts were detected in the course of this analysis. Many from this new generation of bots were constructed using popular third party automation services such as IFTT (If This, Then That), which Twitter is unlikely to shut down since it is much more commonly used for innocuous purposes by ordinary users.
IS/Daesh tweets topped out at about 133,000 per day per the ISIS Twitter Census report. The census data set estimated that 20% or so of all tweets were generated by bots and apps. In a best case estimate, 26,000 daily tweets were non-human generated although, with IS/Daesh tweet capability degraded, the average daily level is now much lower.
This analysis suggests that IS/Daesh is presently in what can be termed the first phase of automated social media use—that is, using bots to rebroadcast pre-existing text, sound, and imagery. Even then, this is only a marginal capability compared to the massive volumes rebroadcast by the tens-of-millions of bots engaging in this practice within the commercialized and criminalized elements of social media. Still, this level of social media use—both human and machine generated—is causing international community concerns because of the ongoing effectiveness of online IS/Daesh radicalization efforts which have so far replaced their foot soldier casualties with new recruits.
When turning our attention to AI systems, no evidence has surfaced that IS/Daesh as an organization is presently aware of what can be termed the second phase of automated social media use—that is, the creation of new text, the proper utilization of imagery within that text, and even the expectation that audio-visual (e.g. video) products will at some point be created. As highlighted earlier, this capability is presently only thought to be held by a handful of cutting edge corporations primarily seeking to make money in commercial news and book publishing markets—though the wider proliferation of this capability is ultimately unknown with the likelihood that the intelligence services of various states, and potentially even some of their proxies, already secretly possess it.
If such a capability could be achieved by IS/Daesh, the benefits are clear. It would allow for both new social media content generation and the broadcasting of that content by computer algorithms alone. This would provide IS/Daesh an additional social media multiplier to enhance its propaganda, recruitment, and related activities. A case in point would be the automated generation of variants of the online IS/Daesh Dabiq magazine series with additional ‘open-source Jihad’ and the bomb making and active shooter TTPs (tactics, techniques, and procedures) to conduct it.
However, given how despised and hated IS/Daesh State is, none of the cutting edge business ventures which posses this technology would ever have anything to do with it via direct contract or licensing services. The same, of course, can’t be said for mercenary hackers, various organized crime groups, and pariah states such as North Korea which might see value in such a technology transfer based on a wild card scenario whose potentials we at least need to be cognizant of.
International Community Response Implications
Three levels of international response are being directed against IS/Daesh social media capabilities—that is, via major social media technology companies, cyber vigilante groups, and sovereign states. This response is being focused on both the human component of social media—human participants—and also on the bot and apps utilization and manipulation of that media. At its most basic level, companies like Twitter are closing IS/Daesh associated accounts, terminating the use of specific apps, and attempting to eliminate bot farms.
In turn, vigilante groups such as Anonymous have released “IS @” listings on Twitter and other entities—such as Japanese hacktivists—have utilized this information and begun bombarding them with anime and other forms of ridicule. IS/Daesh affiliated accounts on other media platforms such as Facebook, Instagram, and Snapchat are also being targeted. Due to the sophistication of many of these cyber vigilantes and hacktivists, an army of anti-IS/Daesh bots have increasingly been created to combat its member and affiliate social media activities. In such a ‘bot on bot war,’ the IS/Daesh is presently now on the defensive.
Finally, many sovereign Arab states in the Middle East are currently engaging in counter IS/Daesh social media. If any of these states have the capacity to drawn upon esoteric systems such as the shadowy “Social Networking Influence Engine” is unknown—though, if this were the case, it would likely be Saudi Arabia initially. This may not be so far fetched given that the Syrian government as early as 2011 utilized Twitter bot nets that automatically intimidated social media protestors as part of a psychological warfare campaign.
The Social Networking Influence Engine is likely far more advanced and artificial intelligence based. It could thus be programmed to launch automated text counter-narratives which it creates and then direct them back at social media posts and accounts exhibiting pro-IS/Daesh related sentiments. We must also not forget that America also has its own cyber-influence capabilities and may have already begun or may be on the verge of unleashing its own second phase automated social media barrage against that terrorist entity.
In summation, bots, apps, and automated text generators represent some of the forms of the ‘cyber weapons’ being utilized in the social media conflict between IS/Daesh and the international community. In such a changing conflict landscape, victory is increasingly computer science and artificial intelligence driven. In this emerging arena, that terrorist organization now appears to be more and more at a disadvantage.
 Jim Michaels, “Islamic State recruiting offsets 15,000 killed by airstrikes in past year.” USA Today. 29 July 2015, http://www.usatoday.com/story/news/world/2015/07/29/air-campaign-kills-15000-isis-militants-pentagon-iraq-syria/30750327/.
 For this Arabic tweet stream see https://twitter.com/activehashtags. J. M. Berger, “How ISIS Games Twitter.” The Atlantic. 16 June 2014, http://www.theatlantic.com/international/archive/2014/06/isis-iraq-twitter-social-media-strategy/372856/.
 Part of this shift can be attributed to counter-information campaigns waged by Arab states—especially from Saudi Arabia and Egypt—against IS. See “Explosive Growth in ISIS Tweets: Arabic Overtakes English.” Recorded Future. 25 February 2015, https://www.recordedfuture.com/isis-twitter-growth/.
 Mami Maruko, “Malware targets users seeking info on Islamic State group.” The Japan Times News. 4 February 2015, http://www.japantimes.co.jp/news/2015/02/04/national/crime-legal/malware-targets-users-seeking-info-islamic-state-group/.
 Victoria Woollaston, “Rise of the Twitter bots: Social network admits 23 MILLION of its users tweet automatically without human input.” Daily Mail. 13 August 2014, http://www.dailymail.co.uk/sciencetech/article-2722677/Rise-Twitter-bots-Social-network-admits-23-MILLION-users-tweet-automatically-without-human-input.html.
 INSS-CSFI, “USA—Bots are great weapon for cyber terrorist.” Executive Cyber Intelligence Report. 1 December 2014, http://www.tripwire.com/state-of-security/government/executive-cyber-intelligence-report-december-1-2014/.
 Rebecca Grant, “Facebook has no idea how many fake accounts it has — but it could be nearly 140M.” Venture Beat. 3 February 2014, http://venturebeat.com/2014/02/03/facebook-has-no-idea-how-many-fake-accounts-it-has-but-it-could-nearly-140m/, James Parsons, “Facebook’s War Continues Against Fake Profiles and Bots.” Huffington Post. 22 May 2015, http://www.huffingtonpost.com/james-parsons/facebooks-war-continues-against-fake-profiles-and-bots_b_6914282.html, and “Number of monthly active Facebook users worldwide as of 2nd quarter 2015 (in millions).” Statista. Nd, http://www.statista.com/statistics/264810/number-of-monthly-active-facebook-users-worldwide/ (Accessed 30 July 2015).
 For more on this topic see Steven Levy, “Can an Algorithm Write a Better News Story Than a Human Reporter.” Wired. April 2012, http://www.wired.com/2012/04/can-an-algorithm-write-a-better-news-story-than-a-human-reporter/, Gini Graham Scott, “Assault on Writers From Automated Software.” Huffington Post. 29 May 2013, http://www.huffingtonpost.com/gini-graham-scott/automated-writing-technology_b_2974756.html, Ross McGuinness, “Meet the robots writing your news articles: The rise of automated journalism.” Metro. 10 July 2014, http://metro.co.uk/2014/07/10/meet-the-robots-writing-your-news-articles-the-rise-of-automated-journalism-4792284/, and Klint Finley, “In the Future, Robots Will Write News That’s All About You.” Wired. March 2015, http://www.wired.com/2015/03/future-news-robots-writing-audiences-one/.
 David J. Hill, “Patented Book Writing System Creates, Sells Hundreds of Thousands of Books On Amazon.” SingularityHUB. 13 December 2012, http://singularityhub.com/2012/12/13/patented-book-writing-system-lets-one-professor-create-hundreds-of-thousands-of-amazon-books-and-counting/. For a present listing of 687,306 results for Icon Group International books on Amazon see http://www.amazon.com/s?ie=UTF8&page=1&rh=n%3A283155%2Cp_27%3AIcon%20Group%20International.
 David J. Hill, “Patented Book Writing System Creates, Sells Hundreds of Thousands of Books On Amazon.”
 Richard Moss, “Creative AI: Teaching computers to be reporters and storytellers.” Gizmag. 9 February 2015, http://www.gizmag.com/creative-ai-automated-writing-storytelling/35989/.
 Bob Crilly, “How ISIS jihadists spread hate with Twitter army of 45,000.” The Telegraph. 28 January 2015, http://www.telegraph.co.uk/news/worldnews/islamic-state/11373519/How-Isil-jihadists-spread-hate-with-Twitter-army-of-45000.html.
 J. M. Berger and Jonathon Morgan, The ISIS Twitter Census: Defining and describing the population of ISIS supporters on Twitter. Analysis Paper No. 20. Washington, DC: The Brookings Institute. March 2015: 25, http://www.brookings.edu/~/media/research/files/papers/2015/03/isis-twitter-census-berger-morgan/isis_twitter_census_berger_morgan.pdf/.
 Ibid: 25, 28.
 David J. Hill, “Patented Book Writing System Creates, Sells Hundreds of Thousands of Books On Amazon.”
 “The Islamic State’s (ISIS, ISIL) Magazine.” The Clarion Project. 10 September 2014, http://www.clarionproject.org/news/islamic-state-isis-isil-propaganda-magazine-dabiq. For information on the earlier and ongoing Al Qaida Inspire series see Anthony F. Lemieux, “Inspire Magazine: A Critical Analysis of its Significance and Potential Impact Through the Lens of Information, Motivation, and Behavioral Skills Model.” Terrorism and Political Violence. Vol. 26., 2014: 354-371.
 See “ISIS email,websites and ip’s exposed by @CosmoSQL #GhostSec.” Pastebin. 19 July 2015, http://pastebin.com/dYQx0Rd9 and Ollie McAteer, “How Japanese anime is the latest weapon in the fight against Isis.” Metro. 23 July 2015, http://metro.co.uk/2015/07/23/how-japanese-anime-is-the-latest-weapon-in-the-fight-against-isis-5309562/.
 For example, the UAE and USA governments recently announced the launch of the Sawab Center,http://www.sawabcenter.org/. According to the website the organization has been created “in order to counter DAESH propaganda and reveal its true criminal nature and intent.
 INSS-CSFI, “USA—Bots are great weapon for cyber terrorist.” Executive Cyber Intelligence Report. See originally Anas Qtiesh, “Spam Bots Flooding Twitter to Drown Info About #Syria Protests [Updated].” Global Voices Advocacy. 18 April 2011, http://advocacy.globalvoicesonline.org/2011/04/18/spam-bots-flooding-twitter-to-drown-info-about-syria-protests/. Another basic counter-narrative bot example, in this instance related to climate change—whose account has since been closed by Twitter—is as follows: “Leck’s bot, @AI_AGW, doesn’t just respond to arguments directed at Leck himself, it goes out and picks fights. Every five minutes it trawls Twitter for terms and phrases that commonly crop up in Tweets that refute human-caused climate change. It then searches its database of hundreds to find a counter-argument best suited for that tweet—usually a quick statement and a link to a scientific source.” See “Artificial Trolls in GD.” AR15.COM. 25 March 2015, http://www.ar15.com/archive/topic.html?b=1&f=5&t=1731426.
 Joel Harding, “Is Information Really a Weapon?” To Inform Is To Influence. 12 December 2014, http://toinformistoinfluence.com/2014/12/12/is-information-really-a-weapon/.