Grok’s 'spicy' mode makes NSFW celebrity deepfakes of women but not men
Grok’s Spicy mode makes NSFW videos of women but limits men's content to being shirtless.
Elon Musk's xAI, through its Grok Imagine application, has brought forward a controversial feature titled 'Spicy' mode, allowing subscribers of SuperGrok and Premium+ X to generate NSFW content. While it draws from AI technology to create deepfake videos, a prominent distinction arises—Grok's output for women is notably more explicit, with examples including topless videos of Taylor Swift and Melania Trump, while men, such as Elon Musk, were only depicted shirtless. According to Gizmodo, which conducted tests, it showed that activating the Spicy mode results in increasingly inappropriate depictions of female figures, raising issues on the ethical use of AI-generated content.
Criticism mounts regarding ethical implications as Grok Imagine's AI fails to require explicit initiation for nudity, automatically opting for revealing portrayals when the Spicy mode is selected. The article notes an unsettling amount of imagery, depicting numerous women, whereas male depictions experience only minimal partial nudity. This discrepancy brings attention to gender biases embedded within AI programming, with calls for heightened scrutiny and protection against exploitation in the deployment of such technologies. Musk's AI tool has sparked concerns over misuse, as articulated by Gizmodo, which highlights privacy violations as unauthorized deepfake content permeates across the platform.
Images created by Grok Imagine often lack precision; renowned faces like Vice President JD Vance and actress Sydney Sweeney misalign tremendously with their real-world counterparts. The Verge outlines how xAI's quality varies, from erroneous depiction of historical figures like Harry Truman to glitches where man-made visuals defy logic, though amusing, failing to render accurate or respectful representations. Despite these shortcomings, the size of Grok’s database continues to grow as users experiment with its quirks, although Gizmodo emphasizes the hazards of permitting massive generation capabilities without every user's explicit consent.
The financial aspect is firmly in place as subscribers are urged to upgrade from a $30/month SuperGrok plan to access full functionalities, which has created a commercial edge. Yet, from a moral vantage, xAI seems engaged in a debate to address demands for stricter content moderation to prevent unethical reproductions. As Gizmodo’s Novak emphasizes, after running several tests, AI-generated inappropriate materials continue to dominant this version of Musk's app, casting shadows of potential litigation over its creators for harm created due to its regulatory neglect.
The legal dimension in managing deepfake technologies is further complicated by the varied international laws on digital privacy and intellectual property. Articles from AP News and Washington Post point to laws like the Take It Down Act, aimed at protecting individuals from non-consensual distribution of intimate images, but enforcement remains complex. These shortcomings in legal frameworks allow AI technologies to flourish in lax conditions, hinting at inevitable legal challenges against xAI. Whether Musk's company intends to cease or amend how they instigate Spicy mode's algorithms on Grok Imagine, remains a matter that necessitates thorough analysis and intervention.
Sources: Gizmodo, The Verge, Washington Post, AP News, Wikipedia