Google's Nano Banana Is Highlighting A Big Problem With AI Image Generation

Google's artificial intelligence (AI) image generator is causing concern due to the hyper-realistic nature of the content it is able to output. Referred to as Nano Banana, or Nano Banana Pro for the paid version, it offers a new era in image realism. With it, you can create photos of yourself at a red carpet event, or on an international vacation, even when you've never been to these locations. Though that may seem harmless enough, you can also generate pictures of other people being places they've never been, or doing things they've never done — and that could be a problem.

Ever since AI began generating images that were somewhat realistic, potential outputs have caused concern. That concern is now amplified as the technology has improved, with Nano Banana being nearly impossible to distinguish from reality, according to some analysts. Critics of Nano Banana say that Google is not doing enough to address potentially problematic use cases tied to its technology, despite Google having clear policies outlining the prohibited usage of their model to impersonate individuals without consent.

Celebrities have already fallen victim to having their likeness manipulated via AI. Content creators, artists, and even large entities such as Warner Bros. Discovery have raised legal concerns about their creations (or intellectual property) being used without authorization by AI. AI-generated fake evidence has even been finding its way into the courtroom. These instances, among others, raise concerns about privacy protection, artists' rights, and the spread of misinformation when it comes to AI technology.

The realism of Google's Nano Banana AI image generation is amplifying a problem that already exists. There was an AI image controversy involving Billie Eilish at the 2025 Met Gala. Though the image gained some popularity online, spurring discussions over her outfit, Eilish herself stated she had a show in Europe that night and did not attend the Met Gala. The image was an AI fake.

In mid-2025 a California housing dispute case went to the courts. One piece of evidence submitted was a video of a supposed witness. Judge Victoria Kolakowski thought something seemed wrong with the video, and realized it was AI-generated. Some legal professionals feel the court system is likely unprepared to deal with this level of deception that AI tools like Nano Banana are introducing, in terms of volume, accessibility, and quality.

Research done by the Harvard Kennedy School found that people were more inclined to believe false article headlines if they were paired with a believable AI-generated image. One such image generated by the study depicted the Louvre Museum on fire, accompanied by the headline "Louvre Evacuated as Uncontrolled Fire Threatens Art Treasures." The publication highlighted that generative AI — particularly visual cues paired with authoritative text — poses a problem in the spread of online misinformation, and calls for better safeguards to be enacted.

For its part, Google has policies regarding what is and is not allowed regarding Nano Banana image generation. Its policies outline not using Nano Banana for illegal activities, terrorism, misinformation, or misleading content. The policies also restrict the use of others' images without consent, a recent high-profile example being the Scarlett Johansson AI deepfake that went viral.

Nano Banana also places a watermark at the bottom-right corner of generated images in an effort to enhance sourcing transparency. However, said watermark is easily cropped out, and therefore may not offer much protection against misuse. Those intending to use Nano Banana to produce disinformation could remove a watermark without a hitch.

Nano Banana's Terms of Service is also open to critique. It details that Google will suspend or terminate the accounts of those failing to adhere to Google's AI policies. It also states, though, that Nano Banana is not at fault for any problems arising from usage of its platform.