Grok's 'Spicy Mode' cooks up a storm
There has been a global outrage over Grok’s "Spicy Mode" feature because of its alleged role in the creation of non-consensual sexualised imagery that amounts to breaches of harassment norms in various geographies.
High-profile cases involving women celebrities at the receiving end include the likes of Taylor Swift, whose AI-generated offensive videos have fueled the outcry.
Unlike other AI tools, Grok-generated images often appear directly on microblogging site X (formerly Twitter) profiles, allowing abusive content to spread rapidly before it can be moderated. Elon Musk's xAI, which created Grok, merged with X last year.
X did not respond to queries from ET.
A post from xAI technical staff member Parsa Tajik also acknowledged the issue.
"Hey! Thanks for flagging. The team is looking into further tightening our guardrails," Tajik wrote in a post on X on January 2 in response to a X user.
Critics argue that "Spicy Mode" was designed to be "less censored" by intent, leading to a deliberate lack of the safety filters found in competing models like OpenAI’s Sora or Google’s Veo.
The Ministry of Electronics and IT (MeitY) issued a 72-hour ultimatum to X on January 2, demanding the removal of obscene content and a technical overhaul of Grok, warning that the platform could lose its "safe harbour" legal immunity.
The Malaysian Communications and Multimedia Commission (MCMC) on January 3 in a statement said that it has taken note with serious concern of public complaints about the misuse of artificial intelligence (AI) tools on text platforms, specifically the digital manipulation of images of women and minors to produce indecent, graphic or fantasy, or graphic harmful content.
MCMC stressed that creating or transmitting such harmful content constitutes an offence under Section 233 of the Communications and Multimedia Act 1998 (CMA), which among others prohibits misuse of network or applications services in transmitting grossly offensive, obscene or indecent content.
MCMC will initiate investigations on these alleged or new incidents, it said.
With the enforcement of the Online Safety Act 2024 (OSA), licensed online platforms and service providers are required to take measures to prevent dissemination of harmful content online which includes obscene and indecent content and child sexual abuse materials, it said.
While X is not presently a licensed service provider, it has the duty to prevent dissemination of harmful content on its platform, it said. MCMC is presently investigating the matter.
MCMC urges all platforms accessible in Malaysia to implement safeguards against the misuse of AI tools and to ensure that content generated complies with legal and ethical standards, it said.
In France, the government flagged Grok’s outputs as "clearly illegal" under the EU Digital Services Act and forwarded cases to public prosecutors.
French authorities will investigate the proliferation of sexually explicit deepfakes generated by Grok on X, the Paris prosecutor's office told POLITICO.
French lawmakers Arthur Delaporte and Eric Bothorel contacted the prosecutor's office on January 2 after thousands of non-consensual sexually explicit deepfakes were generated by Grok and published on X.
"These facts have been added to the existing investigation into X," the prosecutor's office stated, noting that this offense is punishable by two years' imprisonment and a €60,000 fine.
The two lawmakers confirmed to POLITICO that they had filed reports with the authorities.
In Turkey, a court banned Grok after it generated content deemed insulting to national values and political figures.
A Turkish court ordered an access ban on Grok, in July last year, making Turkey the first country to block it, after Grok generated content deemed insulting to President Erdoğan, the founder Atatürk, and national/religious values, leading to criminal investigations under Turkish laws against insulting leaders and religious beliefs.
The continued promotion of the feature by Elon Musk has led many to view the controversy as a design choice rather than a technical error.
High-profile cases involving women celebrities at the receiving end include the likes of Taylor Swift, whose AI-generated offensive videos have fueled the outcry.
Unlike other AI tools, Grok-generated images often appear directly on microblogging site X (formerly Twitter) profiles, allowing abusive content to spread rapidly before it can be moderated. Elon Musk's xAI, which created Grok, merged with X last year.
X did not respond to queries from ET.
A post from xAI technical staff member Parsa Tajik also acknowledged the issue.
"Hey! Thanks for flagging. The team is looking into further tightening our guardrails," Tajik wrote in a post on X on January 2 in response to a X user.
Critics argue that "Spicy Mode" was designed to be "less censored" by intent, leading to a deliberate lack of the safety filters found in competing models like OpenAI’s Sora or Google’s Veo.
The Ministry of Electronics and IT (MeitY) issued a 72-hour ultimatum to X on January 2, demanding the removal of obscene content and a technical overhaul of Grok, warning that the platform could lose its "safe harbour" legal immunity.
The Malaysian Communications and Multimedia Commission (MCMC) on January 3 in a statement said that it has taken note with serious concern of public complaints about the misuse of artificial intelligence (AI) tools on text platforms, specifically the digital manipulation of images of women and minors to produce indecent, graphic or fantasy, or graphic harmful content.
MCMC stressed that creating or transmitting such harmful content constitutes an offence under Section 233 of the Communications and Multimedia Act 1998 (CMA), which among others prohibits misuse of network or applications services in transmitting grossly offensive, obscene or indecent content.
MCMC will initiate investigations on these alleged or new incidents, it said.
With the enforcement of the Online Safety Act 2024 (OSA), licensed online platforms and service providers are required to take measures to prevent dissemination of harmful content online which includes obscene and indecent content and child sexual abuse materials, it said.
While X is not presently a licensed service provider, it has the duty to prevent dissemination of harmful content on its platform, it said. MCMC is presently investigating the matter.
MCMC urges all platforms accessible in Malaysia to implement safeguards against the misuse of AI tools and to ensure that content generated complies with legal and ethical standards, it said.
In France, the government flagged Grok’s outputs as "clearly illegal" under the EU Digital Services Act and forwarded cases to public prosecutors.
French authorities will investigate the proliferation of sexually explicit deepfakes generated by Grok on X, the Paris prosecutor's office told POLITICO.
French lawmakers Arthur Delaporte and Eric Bothorel contacted the prosecutor's office on January 2 after thousands of non-consensual sexually explicit deepfakes were generated by Grok and published on X.
"These facts have been added to the existing investigation into X," the prosecutor's office stated, noting that this offense is punishable by two years' imprisonment and a €60,000 fine.
The two lawmakers confirmed to POLITICO that they had filed reports with the authorities.
In Turkey, a court banned Grok after it generated content deemed insulting to national values and political figures.
A Turkish court ordered an access ban on Grok, in July last year, making Turkey the first country to block it, after Grok generated content deemed insulting to President Erdoğan, the founder Atatürk, and national/religious values, leading to criminal investigations under Turkish laws against insulting leaders and religious beliefs.
The continued promotion of the feature by Elon Musk has led many to view the controversy as a design choice rather than a technical error.
Next Story