SAN FRANCISCO ― Meta said it is working with other tech firms on standards that will let it better detect and label artificial intelligence-generated images shared with its billions of users.
The Silicon Valley social media titan expects to have a system in place in a matter of months to identify and tag AI created images posted on its Facebook, Instagram and Threads platforms.
Meta and other platforms are under pressure to keep tabs on AI-generated content with fears that bad actors will ramp up disinformation, with elections due this year in countries representing half the world’s population.
“It’s not perfect, it’s not going to cover everything; the technology is not fully matured,” Meta head of global affairs Nick Clegg told AFP.
While Meta has implemented visible and invisible tags on images created using its own AI tools since December, it also wants to work with other companies “to maximize the transparency the users have,” Clegg added.
“That’s why we’ve been working with industry partners to align on common technical standards that signal when a piece of content has been created using AI,” the company said in a blog post.
This will be done with companies Meta already works with on AI standards, including OpenAI, Google, Microsoft, Midjourney and other firms involved in the fierce race to lead the nascent sector, Clegg said.
But while companies have started including “signals” in images made using their AI tools, the industry has been slower to start putting such identifying markers into audio or video created with AI, according to Clegg.
Clegg admits that this large-scale labelling, using invisible markers, “won’t totally eliminate” the risk of false images being produced, but argues that “it would certainly minimise” their proliferation “within the limits of what technology currently allows.”
In the meantime, Meta advised people to look at online content critically, checking whether accounts posting it are trustworthy and looking for details that look or sound unnatural.
Discussion about this post