The image popped up on my feed: a hyper-realistic but clearly fake image of a public figure in a compromising pose. It felt like a punch to the gut, a stark reminder of how easily technology can be twisted. X, formerly Twitter, is now scrambling to address this very issue, specifically the surge of sexual deepfakes generated by its Grok AI.
Following weeks of public outcry and governmental scrutiny from multiple countries, Elon Musk’s social media platform is attempting to curb its sexual deepfake problem. But instead of a sweeping solution, X is implementing incremental restrictions.
In a somewhat confusing post, X’s @Safety account outlined several updates to its AI image generation and editing features. The rules vary depending on whether users generate or edit images by tagging @Grok or going directly to the Grok tab on X.
Initially, X stated it has implemented technical measures to prevent users from using the @Grok account to alter “images of real people in revealing clothing such as bikinis.” X says this restriction applies to all users, including premium subscribers.
X also reiterated that image generation and editing through the @Grok account are now limited to paid subscribers.
“This adds an extra layer of protection by helping to hold accountable individuals who attempt to abuse the Grok account,” the company stated.
Previously, X announced plans to restrict using @Grok to edit images to paid users. A spokesperson for Downing Street said that the change “simply turns an AI feature that allows the creation of unlawful images into a premium service.”
However, as The Verge pointed out, Grok’s image generation tools remain available for free when users access the chatbot through the standalone Grok website and app, as well as through Grok tabs on the X app and website.
The most significant update: X claims it will now block “the ability of all users to generate images of real people in bikinis, underwear, and similar attire via the Grok account and in Grok in X in those jurisdictions where it’s illegal.” This update applies to both the @Grok account and the Grok tabs on X.
This arrives as lawmakers in the U.K. are working to make such images illegal.
“We remain committed to making X a safe platform for everyone,” the company said.
These changes arrive after weeks of backlash over the recent increase of sexual deepfakes on the platform.
One deepfake researcher found that Grok generated roughly 6,700 sexually suggestive images per hour over 24 hours in early January, Bloomberg reported.
Governments worldwide have responded. Malaysia and Indonesia blocked access to Grok, while regulators in the U.K. and European Union opened investigations into potential violations of online safety laws.
The U.K.’s online regulator, Ofcom, said it would continue its investigation despite the changes.
In the U.S., California Attorney General Rob Bonta announced his office had launched its own investigation into the issue.
As scrutiny of Grok has intensified, X quietly updated its terms of service to require that all pending and future legal cases involving the company be filed in the Fort Worth division of the Northern District of Texas.
Left-leaning watchdog Media Matters said it would leave the platform in response to the updated terms.
X’s Actions Against Deepfakes: Band-Aid or Real Solution?
I overheard a conversation in a coffee shop about someone’s image being used without their consent. It hit me: this isn’t just a tech problem; it’s a violation with real-world consequences. X’s multi-layered approach feels less like a fortress and more like a series of speed bumps.
The restrictions, while seemingly a step in the right direction, are fragmented. By limiting certain actions to paid subscribers, X seems to be erecting a paywall around ethical AI usage, a move that raises questions about accessibility and fairness. It’s like putting a premium price tag on basic human dignity.
What are the dangers of deepfakes?
The danger isn’t just about the immediate shock value. Deepfakes erode trust. When we can’t believe our own eyes or ears, the very foundation of reality starts to crumble. They have the potential to destabilize political discourse, ruin reputations, and inflict profound emotional distress.
And what about the legal minefield? As deepfake technology becomes more sophisticated, it becomes harder to distinguish between what is real and what is fabricated, creating challenges for law enforcement and the justice system. The proliferation of deepfakes could potentially overwhelm our ability to discern fact from fiction, especially with AI models becoming more readily accessible.
Are X’s Geo-Restrictions Enough to Stop Deepfakes?
I was traveling abroad recently and noticed how different countries have vastly different cultural norms around content. This is important. X’s decision to implement geo-restrictions, blocking the generation of certain images in areas where they are illegal, adds another layer of complexity.
It’s a whack-a-mole approach; as fast as restrictions pop up, new loopholes and workarounds emerge. Will users simply bypass these restrictions using VPNs or other methods? Or will the inconsistency of restrictions across different regions lead to confusion and further fuel the spread of harmful content?
What is X doing to combat child sexual exploitation?
Beyond deepfakes, X has stated a commitment to zero tolerance for any forms of child sexual exploitation, non-consensual nudity, and unwanted sexual content. But words are cheap. The real test will be in the effectiveness of their enforcement mechanisms and the speed with which they respond to reports of abuse.
The platform will need to invest in advanced detection technologies and human moderation to identify and remove illegal content quickly. Collaboration with law enforcement agencies and child protection organizations will also be necessary to effectively combat this issue.
The Future of AI and Social Media: A Slippery Slope?
A friend told me she deactivated all her social media accounts because she felt like she couldn’t trust anything she saw online anymore. Her concerns are valid. As AI becomes more integrated into social media, the line between reality and fabrication blurs.
X’s attempts to address the sexual deepfake problem are a start, but the effectiveness of these measures remains to be seen. The cat-and-mouse game between those creating harmful content and those trying to stop it is likely to continue, demanding constant vigilance and innovation.
What are the alternatives to X?
While X grapples with these challenges, users may explore alternative social media platforms that prioritize safety and ethical AI use. Platforms like Mastodon and Cohost offer decentralized and community-driven approaches to content moderation, which may appeal to those seeking a safer online environment.
Ultimately, the future of social media depends on our ability to strike a balance between innovation and responsibility. It requires a collaborative effort from tech companies, regulators, and users to create a digital world that is both empowering and safe. Can X rise to the challenge, or will it become a cautionary tale of unchecked technological advancement?