I am writing to formally express my concern about the recent widespread misuse of your image generation tools to create large quantities of insulting avatars depicting political leaders and public figures. This activity constitutes a clear violation of both your published policies and technical commitments, while posing significant risks to social order and individual rights.
1.Policy Violation:
• Per your January 2024 policy update addressing election-year risks, OpenAI explicitly prohibits generating images of real individuals including political candidates and public figures.
• Your November 2024 transparency report confirms technical interventions blocking over 250,000 requests targeting presidential candidates through DALL-E’s filtering mechanisms.
2.Systemic Security Flaws:
• Recent social media trends demonstrate users successfully circumventing safeguards by manipulating prompts to create derogatory content. Examples include:
▫ Cartoon depictions of Donald Trump wearing clown accessories while holding "LOSER" signage
▫ Other politically charged caricatures containing demeaning visual elements
• Such content risks damaging reputations, inciting social confrontation, and contradicts your platform’s ethical guidelines.
3.Required Actions:
• Conduct retrospective audits of generation logs to identify policy violations
• Suspend offending accounts and remove disseminated abusive content
• Enhance real-time filtering protocols for political figure-related prompts
• Implement mandatory request termination for avatar generation targeting public officials
• Strengthen audit mechanisms to prevent future policy circumvention
As an AI leader bearing social responsibility, we urge immediate rectification of these security vulnerabilities and strict enforcement of published content policies. Please provide confirmation of corrective actions.