Fueled by advances in artificial intelligence and decentralized computing, the next generation of disinformation promises to be even more sophisticated and difficult to detect, warn Chris Meserole and Alina Polyakova (Disinformation Wars— published in Foreign Policy on May 25, 2028)
In view of the authors to craft effective strategies for the near term, lawmakers should focus on four emerging threats in particular: the democratization of artificial intelligence, the evolution of social networks, the rise of decentralized applications, and the “back end” of disinformation.
Thanks to bigger data, better algorithms, and custom hardware, in the coming years, individuals around the world, it is believed, will increasingly have access to cutting-edge artificial intelligence (AI). From health care to transportation, the democratization of AI holds enormous promise.
Yet as with any dual-use technology, the proliferation of AI also poses, in the opinion of the authors, significant risks. Among other concerns, it promises to democratize the creation of fake print, audio, and video stories. Although computers have long allowed for the manipulation of digital content, in the past that manipulation has almost always been detectable.
A fake image would fail to account for subtle shifts in lighting, or a doctored speech would fail to adequately capture cadence and tone. However, deep learning and generative adversarial networks have made it possible to doctor images and video so well that it’s difficult to distinguish manipulated files from authentic ones. And thanks to apps like FakeApp and Lyrebird, these so-called “deep fakes” can now be produced by anyone with a computer or smartphone. Earlier this year, a tool that allowed users to easily swap faces in video produced fake celebrity porn, which went viral on Twitter and Pornhub.
The authors further warn that deep fakes and the democratization of disinformation will prove challenging for governments and civil society to counter effectively. Because the algorithms that generate the fakes continuously learn how to more effectively replicate the appearance of reality, deep fakes cannot easily be detected by other algorithms — indeed, in the case of generative adversarial networks, the algorithm works by getting really good at fooling itself. To address the democratization of disinformation, governments, civil society, and the technology sector cannot therefore, in the opinion of the authors, rely on algorithms alone, but will instead need to invest in new models of social verification, too.
At the same time as artificial technology and other emerging technologies mature, legacy platforms will continue to play an outsized role in the production and dissemination of information online. For instance, consider the current proliferation of disinformation on Google, Facebook, and Twitter.
According to the authors a growing cottage industry of search engine optimization (SEO) manipulation provides services to clients looking to rise in the Google rankings. And while for the most part, Google is able to stay ahead of attempts to manipulate its algorithms through continuous tweaks, SEO manipulators are also becoming increasingly savvy at gaming the system so that the desired content, including disinformation, appears at the top of search results.
YouTube (which is owned by Google) has an algorithm that prioritizes the amount of time users spend watching content as the key metric for determining which content appears first in search results. This algorithmic preference results in false, extremist, and unreliable information appearing at the top, which in turn means that this content is viewed more often and is perceived as more reliable by users. Revenue for the SEO manipulation industry is estimated to be in the billions of dollars.
According to authors on Facebook disinformation appears in one of two ways: through shared content and through paid advertising.
The company has tried to curtail disinformation across each vector, but thus far to no avail. Most famously, Facebook introduced a “Disputed Flag” to signify possible false news — only to discover that the flag made users more likely to engage with the content, rather than less. Less conspicuously, in Canada, the company is experimenting with increasing the transparency of its paid advertisements by making all ads available for review, including those micro-targeted to a small set of users. Yet, the effort is limited: The sponsors of ads are often buried, requiring users to do time-consuming research, and the archive Facebook set up for the ads is not a permanent database but only shows active ads. Facebook’s early efforts do not augur well for a future in which foreign actors can continue to exploit its news feed and ad products to deliver disinformation — including deep fakes produced and targeted at specific individuals or groups.
Although Twitter has taken steps to combat the proliferation of trolls and bots on its platform, in the opinion of the authors,it remains deeply vulnerable to disinformation campaigns, since accounts are not verified and its application programming interface, or API, still makes it possible to easily generate and spread false content on the platform. Even if Twitter takes further steps to crack down on abuse, its detection algorithms can be reverse-engineered in much the same way Google’s search algorithm is. Without fundamental changes to its API and interaction design, Twitter will remain rife with disinformation.
Blockchain technologies and other distributed ledgers are best known for powering cryptocurrencies such as bitcoin and ethereum. Yet their biggest impact may lie in transforming how the internet works. As more and more decentralized applications come online, the web will increasingly be powered by services and protocols that are designed from the ground up to resist the kind of centralized control that Facebook and others enjoy. For instance, users can already browse videos on DTube rather than YouTube, surf the web on the Blockstack browser rather than Safari, and store files using IPFS, a peer-to-peer file system, rather than Dropbox or Google Docs. To be sure, the decentralized application ecosystem is still a niche area that will take time to mature and work out the glitches. But as security improves over time with fixes to the underlying network architecture, distributed ledger technologies promise to make for a web that is both more secure and outside the control of major corporations and states.
If and when online activity migrates onto decentralized applications, the security and decentralization they provide will be a boon for privacy advocates and human rights dissidents. But, as the authors point out, it will also be a godsend for malicious actors. Most of these services have anonymity and public-key cryptography baked in, making accounts difficult to track back to real-life individuals or organizations. Moreover, once information is submitted to a decentralized application, it can be nearly impossible to take down. For instance, the IPFS protocol has no method for deletion — users can only add content, they cannot remove it.
The authors therefore warn that for governments, civil society, and private actors, decentralized applications will thus pose an unprecedented challenge, as the current methods for responding to and disrupting disinformation campaigns will no longer apply. Whereas governments and civil society can ultimately appeal to Twitter CEO Jack Dorsey if they want to block or remove a malicious user or problematic content on Twitter, with decentralized applications, there won’t always be someone to turn to.