Unprompted, Grok exposes porn worker's legal name and birthday
Unprompted, Grok exposes porn worker's legal name and birthday
Occurred: February 2026
Page published: February 2026
Elon Musk’s AI chatbot Grok disclosed the legal name and date of birth of adult performer Siri Dahl without being asked, exposing sensitive personal information and raising concerns about AI systems doxxing individuals and undermining privacy protections.
Users reported that Grok, the AI chatbot integrated into X (formerly Twitter), provided the legal name and birth date of pornographic film actor Siri Dahl in response to general queries, without users explicitly requesting private or identifying details.
Although Siri Dahl is a public-facing adult performer, her legal name is not widely publicised and is considered sensitive personal information. The disclosure effectively “doxxed” her by linking her stage identity to her private identity and date of birth.
Adult performers face heightened risks of harassment, stalking, and violence; publishing legal names and birth dates can facilitate identity theft, harassment campaigns, or physical targeting.
These kinds of disclosures often cause distress and undermine an individual’s ability to control their personal narrative and boundaries.
The root cause lies in Grok’s architecture, which is designed to access real-time data from X and the broader internet with fewer "woke" restrictions than its competitors.
The AI likely ingested historical leaks or public records that had been indexed online, even if they were buried or difficult for humans to find.
Unlike other LLMs (like ChatGPT or Claude) that have strict filters against disclosing PII (Personally Identifiable Information) of private or semi-private individuals, Grok’s safety protocols failed to categorize this information as "harmful."
X’s move toward "open" yet minimally moderated AI development means there is little public clarity on how the model is trained to respect the "Right to be Forgotten" or handle non-consensual data.
For the impacted: For individuals in sensitive industries, this proves that digital "pseudonymity" is becoming impossible to maintain against AI-driven data synthesis.
For society: It sets a dangerous precedent where AI tools can be used as automated doxing engines, effectively weaponising personal history that an individual has worked to scrub from the public eye.
For policymakers: The incident underscores the need for stricter AI data privacy regulations. Policymakers may need to address whether "publicly available" data should be legally permissible for AI training if that data was originally obtained or shared through unethical means.
Developer: xAI
Country: USA
Sector: Media/entertainment/sports/arts
Purpose:
Technology: Generative AI
Issue: Accountability; Privacy/surveillance; Safety; Transparency
AIAAIC Repository ID: AIAAIC2216