It's never been easier to AI a Thanksgiving dinner table — just look at social media (www.businessinsider.com)

Over the Thanksgiving holiday social media filled with AI-generated family portraits and staged celebrity dinners — many shared as jokes by public figures like RFK Jr., Alex Jones, crypto influencer Tiffany Fong and tech exec Daniel Newman (who posted eerily realistic images of himself with CEOs such as Mark Zuckerberg, Elon Musk and Tim Cook). The surge followed the launch of Google Gemini’s Nano Banana Pro a week earlier; users comparing results to the prior Nano Banana flagged a noticeable jump in photorealism. Other generative tools also made holiday appearances (OpenAI’s Sora produced animated turkey clips; Topaz Labs restored vintage parade footage), but Nano Banana Pro dominated discussion for how convincingly it can render fabricated scenes. The takeaway for the AI/ML community is clear: generative-image models are rapidly closing the gap between playful fakery and credible deepfakes. Technical upgrades in Nano Banana Pro show tangible improvements in realism and compositing, suggesting that images which are currently humorous could become indistinguishable from real photos within a short time frame — raising stakes for provenance, detection, and platform policy. The episode underscores a near-term need for better watermarking, robust detection classifiers, and media literacy as generative models improve both creative uses and potential misuse.