Your video classifier is a masterpiece of creative destruction
Adobe Firefly Team,
I'm writing to formally congratulate you on building the most effective anti-creativity tool in the history of digital media.
In the past week, your classifier has blocked:
- A woman walking forward with a serious expression
- A woman standing in a corridor
- A cyberpunk action scene with armored guards and energy effects
No nudity. No gore. No weapons aimed at real people. No hate speech. No copyrighted characters. Just women existing in sci-fi environments, which your system apparently classifies as a threat to civilization.
I'm a looking for professional who creates artistic video content. I use multiple AI platforms daily; Kling, Seedance, Suno, Midjourney, and none of them have ever blocked a woman walking. Your platform is the only one that treats basic human locomotion as a policy violation.
The irony is exquisite: I'm using Adobe, the company that built Premiere Pro and After Effects, tools that have edited actual war footage, horror films, and every violent blockbuster of the last 30 years, and your generative AI won't let me animate a woman in a dark hallway.
Your classifier doesn't protect anyone. It doesn't prevent harm. What it does, with remarkable efficiency, is punish uswers (even paid ones) for attempting to use your product for its intended purpose. Every false positive is a customer who subscribe the REAL Kling instead. Every blocked prompt is revenue walking out the door. Every "Possible Policy Violation" popup on an innocuous scene is a middle finger to the creative professionals who trusted your brand.
I understand legal caution. I understand liability management. What I don't understand is how a company with your resources can't build a classifier that distinguishes between actual harmful content and a woman in a sci-fi costume standing in a building.
Fix your classifier or rename the product. "Adobe Firefly" suggests creative spark. "Adobe Fire Extinguisher" would be more accurate.
Regards,
Someone that will NEVER BE YOUR CUSTOMER
I'm writing to formally congratulate you on building the most effective anti-creativity tool in the history of digital media.
In the past week, your classifier has blocked:
- A woman walking forward with a serious expression
- A woman standing in a corridor
- A cyberpunk action scene with armored guards and energy effects
No nudity. No gore. No weapons aimed at real people. No hate speech. No copyrighted characters. Just women existing in sci-fi environments, which your system apparently classifies as a threat to civilization.
I'm a looking for professional who creates artistic video content. I use multiple AI platforms daily; Kling, Seedance, Suno, Midjourney, and none of them have ever blocked a woman walking. Your platform is the only one that treats basic human locomotion as a policy violation.
The irony is exquisite: I'm using Adobe, the company that built Premiere Pro and After Effects, tools that have edited actual war footage, horror films, and every violent blockbuster of the last 30 years, and your generative AI won't let me animate a woman in a dark hallway.
Your classifier doesn't protect anyone. It doesn't prevent harm. What it does, with remarkable efficiency, is punish uswers (even paid ones) for attempting to use your product for its intended purpose. Every false positive is a customer who subscribe the REAL Kling instead. Every blocked prompt is revenue walking out the door. Every "Possible Policy Violation" popup on an innocuous scene is a middle finger to the creative professionals who trusted your brand.
I understand legal caution. I understand liability management. What I don't understand is how a company with your resources can't build a classifier that distinguishes between actual harmful content and a woman in a sci-fi costume standing in a building.
Fix your classifier or rename the product. "Adobe Firefly" suggests creative spark. "Adobe Fire Extinguisher" would be more accurate.
Regards,
Someone that will NEVER BE YOUR CUSTOMER
AdChoices