Generative artificial intelligence (AI) is fundamentally altering our relationship with visual media, making it increasingly difficult to distinguish between real and fabricated images. This shift poses significant risks to public trust and democratic discourse, as images historically serve as powerful evidence of events. The article highlights the urgent need for a comprehensive social and legal framework to address these challenges. Proposed solutions include mandatory labeling of AI-generated images and the implementation of cryptographic watermarking to trace the origins and alterations of visual content. Such measures would help restore credibility to images, which have historically played a crucial role in shaping public understanding and political will. The phenomenon known as the 'liar's dividend' complicates matters further, as it allows malicious actors to undermine genuine images by labeling them as fakes. The author argues that while technology alone cannot solve this issue, a combination of legal accountability, technical innovations, and institutional support can help reestablish trust in visual media. Ultimately, the goal is not to return to a time of naive visual trust but to create robust mechanisms that adapt to the realities of digital manipulation, ensuring that images continue to anchor shared understanding in society.