Q1: What legal protections currently exist against non-consensual AI imagery in the United States?
A1: Victims currently rely on a patchwork of state-level criminal and civil statutes in states such as California, New York, and Minnesota. At the federal level, bipartisan bills including the DEFIANCE Act seek to establish direct civil remedies enabling victims to sue perpetrators for significant statutory damages.
Q2: Why are platforms slow to remove AI-generated imagery?
A2: Traditional detection tools rely on static hash matching, which struggles to detect variations of synthetically generated content. Furthermore, bad actors intentionally modify compression, framing, and filenames to bypass automated trust and safety filters.
Q3: How do open-source tools contribute to the issue?
A3: While commercial tools hosted by major tech firms implement strict safety filters, open-weight image models run locally without centralized guardrails, allowing users to train and deploy specialized likeness models without oversight.
