President Donald Trump shared a series of images on Truth Social claiming to show White House renovation progress. One "before and after" image of a column drew scrutiny after Reddit users identified multiple visual inconsistencies. The "restored" column displays an uneven number of steps on either side, and the step count does not align with the "before" version. The two photographs appear shot from the same angle and preserve identical window reflections. Protective cardboard at the column's base appears in the same position between both shots. The "newly-painted" surface looks unnaturally smooth, lacking the texture of real paint work.
The Verge confirmed that the image posted to Truth Social triggers Google's SynthID Detector, which scans for an invisible watermark indicating creation or editing with AI tools. The system labeled the file as "made or edited with OpenAI." The SynthID watermark did not appear on other Trump photographs, including ones of the White House facade and heliport. The White House did not respond to a request for comment. Photographs from June show workers painting the exterior, suggesting a fresh coat may have been applied, but the AI-modified image contains fabricated details.
Reddit users flag step mismatch in Trump image
Community members on Reddit were the first to document the discrepancies. The "restored" column shows steps that are not level across both sides. The step configuration in the "before" photograph does not correspond to the altered version. In addition to the step count mismatch, the before and after frames preserve nearly identical lighting, camera angle, and window reflections. The protective cardboard shielding the column's base remains in the same position in both frames, suggesting the image was composited rather than captured as a true before-and-after pair.
Google SynthID detector flags OpenAI edit
Google's SynthID Detector identified the image as generated or modified with OpenAI technology. SynthID embeds an invisible watermark into AI-created media, and the system flagged the Trump column image as matching that pattern. The detector did not activate on the President's other Truth Social posts showing the White House exterior and heliport. This contrast indicates the column image may have undergone post-processing with an AI image generator, while the other photographs were taken conventionally.
Trump's history with AI tools
The President has a pattern of posting AI-generated content. He regularly shares AI-created caricatures of himself. The White House operation has been known to distribute racist AI memes and at least one deceptively edited photograph that made a protester appear distressed. Those prior instances have drawn criticism from watchdog groups and journalists tracking the spread of synthetic media during the 2024 election cycle.
Real photographs show painting in June
Unaltered photographs from June document workers painting the White House facade. Those images show the column and surrounding structure under normal conditions, without the exaggerated smoothness or step inconsistencies present in the Trump-published version. The timing suggests the column may have received a fresh coat of paint, but the AI-modified image distorts the actual changes for a political purpose.
Implications for social media and politics
The incident adds to a growing record of synthetic media used by political figures to shape public perception. AI-edited images can obscure factual details, making it harder for audiences to distinguish between real events and fabricated narratives. The ease with which Google's detector identified the watermark highlights both the presence of embedded markers in some AI outputs and the need for viewers to apply critical scrutiny when consuming political content shared on platforms like Truth Social.