In 2015, Elon Musk and Sam Altman founded OpenAI with the noble aim of advancing AI for the benefit of humanity. Fast forward a decade, and Musk’s new venture, xAI, finds itself embroiled in controversy with its generative AI tool, Grok AI.
Grok AI is presented as a quirky, rebellious AI assistant, but it has stirred trouble with its lack of guardrails, making it prone to spreading misinformation, such as false details about the 2024 US election. This unrestrained design was already criticized by European regulators in July for auto-enrolling X platform users into its data training programs without explicit consent.
Grok’s image generation capabilities further deepen the concerns. Users have shown how easy it is to create inflammatory depictions of public figures like Kamala Harris and Donald Trump. Meanwhile, Musk is integrating Grok deeply into X, offering real-time data for personalized news feeds and post composition, currently available only to Premium+ subscribers.
According to Camden Woollven from GRC International Group, Grok’s real-time data access allows it to discuss unfolding events, giving it an edge over competitors. Nathan Marlor from Version 1 argues that Grok’s transparency comes at the cost of heightened bias, as it has fewer guardrails than counterparts like OpenAI and Anthropic.
The AI has come under fire for providing incorrect election information, leading to complaints from election officials in several states. The company responded by redirecting election-related queries to Vote.gov, but it continued to urge users to independently verify any received information.
Grok also collects vast amounts of data that raise significant privacy issues. Automatically opted in, X users’ posts and interactions contribute to Grok’s training, which is described in detail on xAI’s Help Center page. Angus Allan from CreateFuture suggests that Musk’s firm may have overlooked consent requirements set by the EU’s GDPR.
To protect personal data, users are advised to utilize X’s privacy settings. Switching an account to private and opting out of model training are effective steps. Even if individuals no longer use X, they should still log in to opt out, ensuring that past posts aren’t used.
Moreover, xAI allows users to delete their conversation history, with a 30-day period before deletion takes effect, barring legal or security reasons. Chief Technology Officer at NordVPN, Marijus Briedis, emphasizes the importance of staying informed about privacy policy updates and managing content shared on X.
With its unfiltered approach, Grok AI offers a fascinating yet concerning glimpse into the future of AI. As it continues to evolve, taking proactive steps to protect personal data and staying informed about privacy measures will be crucial.
Source: Wired