Facebook, Microsoft, Twitter and YouTube (a subsidiary of Google/Alphabet) on Monday announced a global partnership to more quickly identify "terrorist content" on their platforms and thus curb its spread.
Ahe four U.S. companies plan to create a common database of digital "fingerprints" of propaganda and recruitment photos or videos removed from their platforms, based on a common message relayed on their respective sites. "By sharing this information with each other, we can use (these digital fingerprints) to help identify potentially terrorist content on our respective consumer platforms," they explain.
However, no message will be automatically removed or blocked: it will be up to each company to assess whether the identified content violates its own rules. Each company will also decide independently which images and videos it decides to add to the common database. They specify that they want to start with the "most extreme and blatant" images and videos removed from their platforms, and therefore "the most likely to violate the rules of all our companies".
Involving other web actors
Microsoft, Facebook, Twitter and YouTube also say they want to consider "how to involve additional companies in the future," but insist that each of them will continue to handle independently requests for information or content removal from governments or law enforcement agencies.
The initiative comes at a time when the United States, the European Commission and a series of other governments have in recent months made repeated calls for social networks to step up their fight against online jihadist propaganda.
Following these calls, Twitter reported that it has suspended more than 360,000 accounts promoting terrorism since mid-2015 and has seen a decline in the use of its platform by jihadists, according to Sinead McSweeney, vice president for public policy for the social network in the Europe/Middle East/Africa region.
In a comment sent to AFP, she said that a large proportion of the account suspensions carried out by Twitter had been detected by technical means such as anti-spam tools, but that the sharing of digital fingerprints with the common database would be done "manually" and "on a periodic basis". There is no one "one-size-fits-all" approach to tackle this type of material; each platform is different," she said.
There's no magic algorithm.
Most social networks prohibit in their rules of use content calling for violence or glorifying terrorism. Even if they have improved their automatic detection tools, industry players regularly reaffirm that they do not have a "magic" algorithm to identify problematic content and rely heavily on their users' reporting of contentious content.
The online distribution of "terrorist" content is "a pressing problem that requires special attention from technology companies", and therefore justifies the joint initiative announced on Monday, noted one of the participating companies.
A never-ending race
Will the measures taken be sufficient to hinder the actions of cyber-jihadists? Nothing is less certain, according to experts. This policy risks pushing users even further towards more confidential, encrypted or protected social networks, or even towards the "darkweb" or "deepweb", an obscure part of the internet not referenced in traditional search engines. Gérôme Billois, an expert with the French Information Security Club (Clusif), said last August: "the very nature of the Internet means that it is an endless race, in which you are always one notch behind". "We must always think strategy, battle, military tactics." told at AFP the rhetorician and philosopher Philippe-Joseph Salazar, author of the essay "Paroles armées - Comprendre et combattre la propagande terroriste" (Lemieux éditeur). "Twitter was a battleground. If this terrain disappears or is less easy, we move the battalions elsewhere, that's all. And then we end up with the problem of Telegram, or Darknet."
American expert Andrew Macpherson, a cybersecurity specialist at the University of New Hampshire, said: "Terrorist groups will certainly continue to use new technologies for propaganda by all means," he added. "They will always be looking for ways to maintain and improve the confidentiality of their communications.
For this purpose, anonymization, encryption and cyber-dissimulation software are easy to find on the web. No special technical skills are required to use them, as recent cases in which investigators have been arrested in their investigations using encrypted telephones, messaging systems protected by unbreakable passwords or private forums that they have not been able to penetrate have shown.