Since the start of the Russian offensive in Ukraine a year ago, pro-Russia influence campaigns have multiplied on social networks, but without much effect, Meta, the parent company of Facebook and Instagram, said on Thursday. .
“
We are seeing an increasing number of attempts by Russian networks to create fake accounts
” to promote the war and spread anti-Ukraine disinformation on the platforms, said Nathaniel Gleicher, head of security policy at Facebook, during a press conference.
“
They try everything and anything, to see what takes
,” he summed up.
“
Unfortunately for them, these accounts have very little following.
Most are automatically detected and removed before they can build an audience
,” he said.
The Californian group regularly takes stock of its fight against manipulation operations on its networks, whether espionage, hacking or disinformation.
Read alsoLaurent Solly (Meta): “We have entered a major cycle of technological breakthrough with AI and the metaverse”
When Russia invaded Ukraine and intensified censorship, Meta took steps to limit the room for maneuver of its state media (even banishing them from Europe, following the demands of the European Union).
The use of generative AI closely monitored
"
Six months after the invasion, we observed an 80% drop in engagement (reading and interaction, editor's note) with the content of these entities
", underlined Nathaniel Gleicher.
As a result, these actors have focused on their "
clandestine activities ", adopting "
spamming
" tactics
rather than "
sophisticated operations like those we have dismantled in the past
", he continued.
The executive also said that the company is closely monitoring the possibility of malicious actors using generative artificial intelligence (AI) to create more effective disinformation campaigns.
The success of the ChatGPT interface of the OpenAI start-up, capable of generating a wide variety of original texts (writings, slogans, emails, etc.), has indeed raised the concern of many experts.
“
We have seen these generative AI techniques being used in operations, in particular to create photos and images to illustrate false accounts
”, acknowledged Nathaniel Gleicher.
“
But our systems track behaviors (of inauthentic users), not content
,” he explained.
"
These accounts behave differently than real people
."