Financial Markets


A brewing issue within Meta's recently introduced AI labelling system has sparked widespread criticism from photographers and creators. Erroneously affixing the “Made by AI” label to original photographs, Meta's system seems to misclassify human creativity as machine algorithm outputs. The impacts of this situation are multifaceted, impacting artistic copyright, content credibility, and drawing a blurry line between human-led creations and AI-powered creations, which may potentially reshape the future of the digital content landscape.

Prominent mislabelling instances have already been noted, such as ex-White House photographer Pete Souza's basketball game picture and the photo of the Kolkata Knight Riders' India Premier League Cricket tournament victory, which were falsely categorized as AI-generated. This issue appears to be more pronounced when viewing images on mobile devices, creating a skewed perception of content origination for users primarily accessing Meta on their smartphones.

Some experts suggest that the use of Adobe’s cropping tool, turning images into JPEG format or employing AI tools like Adobe’s Generative Fill tool to remove small objects could be triggering Meta's AI labelling algorithm. However, this assumption further raises the question of fairness and integrity of Meta's AI algorithms, considering photographers' stance that slight edits shouldn’t translate their human-curated work as AI-generated.

From the creators’ perspective, the erroneous AI label undermines their creative identity and the labor invested in these works, potentially leading to under-appreciation. Photographers have been vocalizing their frustration, highlighting the need for more stringent definitions and regulations around the use of AI labels.

In response to the burgeoning concerns, Meta declared that it’s considering the extensive feedback and reassessing its labelling approach. The aim is to ensure the “Made with AI” label appropriately and accurately reflects the extent of AI utilisation in an image. This self-correction motion will be crucial in ensuring the balance between the integration of AI tools in digital creations and the rightful recognition of human creativity.

Meta's AI label was introduced in February, incorporated photos generated with tools from several companies including Google, OpenAI, Microsoft, Adobe, Midjourney, and Shutterstock. Importantly, the exact conditions prompting the label's usage remain undisclosed, intensifying the concerns about its credibility and accuracy.

Many of these technology giants are reportedly integrating metadata to indicate the use of their AI tools. For instance, Adobe initiated the addition of origin content in metadata via its Content Credentials system launched last year. But as the Meta issue demonstrates, amendments are needed to fine-tune the balance between AI and human contribution identification.

These discussions and the ongoing challenges underpin one critical future impact: the necessity to clearly demarcate the boundaries between human creations and AI-generated content. With the rise of AI in multiple dimensions of our lives, finding an optimal blend of human intelligence and AI, while respecting and acknowledging original human creativity, is a mission-critical task that will shape how we perceive, trust, and value digital content.