Adobe Lied! They Used 1000s Of Midjourney Images To Build Adobe Firefly

Adobe’s Firefly AI image generator, touted as ethically trained on licensed stock images, faces scrutiny due to potential data sourcing issues.

Adobe Lied! They Used 1000s Of Midjourney Images To Build Adobe Firefly

Bloomberg’s report suggests Firefly’s training data may include images from competitor Midjourney, suspected of scraping internet images without licensing.

Adobe refutes claims, stating only 5% of Firefly’s training data may be questionable, insisting all were rigorously moderated from the Adobe Stock library.

Why Adobe Is Scared?

Adobe introduced Firefly with a promise of indemnity against copyright theft claims for its enterprise clients, aiming to assure them of its safety.

Firefly was marketed by Adobe as a secure option compared to other platforms like Midjourney and DALL-E, with all data licensed and cleared for model training.

Some artists felt pressured into allowing their work to be used by Adobe, though Firefly users believed that any image created with it was free from copyright infringement risks.

Is Adobe Firefly’s Images Copyright free?

Despite concerns over image sources, Adobe assures the safety of all non-human images in its stock, affirming rigorous moderation.

A spokesperson stated that every image submitted to Adobe Stock undergoes thorough moderation to avoid copyright infringement or trademark violations.

Adobe appears to be implementing stricter measures in its AI video generator project, reportedly compensating artists per minute for their video contributions.

Related Stories:

Help Someone By Sharing This Article