MIT Research Shows AI Image Attribution Declines with Expanding Training Data

MIT researchers have identified "attribution decay" in generative AI, where tracing an AI-generated image to its source becomes harder as training datasets grow. This finding, particularly for diffusion models, could complicate future copyright disputes and regulatory oversight regarding AI-generated content. It highlights a growing challenge in AI accountability.