Advocating Pixel-Level Authentication of Camera-Captured Images
Bibliographic record
Abstract
The authenticity of digital images posted online and shared on social media is often questioned due to the ability of photo-editing software to alter image content and generative AI methods that can produce visually compellingdeepfakes. Only images directly produced by cameras are deemed unaltered and beyond suspicion, as they have not undergone any modifications. However, there is a recent trend among camera manufacturers to integrate AI-based modules into the dedicated onboard hardware, specifically the image signal processor (ISP), responsible for processing the captured sensor image into the final saved image for users. Many of these AI modules utilize perceptual or generative losses during training, which can “hallucinate” image content. While this hallucinated content often manifests as small details and textures, there are instances where these regions unintentionally impact the interpretation of the entire image. This paper aims to bring attention to this issue and advocate for in-camera strategies to validate the authenticity of camera-captured images at a pixel level. We propose the creation of an "authenticity" mask that could be stored as additional metadata with each image. This information can be extracted and overlaid on the image to easily identify the hallucinated regions. Considering the widespread implications of image authenticity (e.g., in courtroom evidence, news broadcasts, and other media forms), we anticipate that authentication metadata will become a standard practice for any ISP utilizing AI.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.013 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.003 | 0.005 |
| Open science | 0.001 | 0.003 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.004 | 0.003 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".