Google's AI Blunder and Fresh Blood for MFT – This Week's Photo News 7/26
Google punishes a photographer for good work, while the Micro Four Thirds system gains a new partner from China.
(Image: Shutterstock/Alexander Supertramp)
Sometimes the brave new tech world feels like a scene from a Franz Kafka novel, only that the inaccessible official in the castle today is an algorithm. And if this algorithm decides that black is actually white, then that's how it is. Objection? Futile. A particularly absurd episode of this kind was experienced and meticulously documented by journalist and photographer Georg Berg this week. It's a story that should make anyone who picks up a camera sit up and take notice.
Google's “Won't Fix” Policy: When Good Work Becomes Suspected AI
The case is so paradoxical that it could hardly be invented: Georg Berg took an authentic reportage photo – four elderly gentlemen in traditional Swiss attire. He then professionally edited it in Lightroom to precisely separate the people from the background. A diligent piece of craftsmanship that every good photographer knows. The result? Google's AI detector “SynthID” stamped the image as “AI-generated”. The justification, if one can call it that, lies in the precision: the clean masking created statistical patterns that the algorithm interpreted as “unnatural” and therefore the work of a machine. A professional who works too well is thus declared a forger.
The real icing on this cake of irony, however, is that the photo possessed a C2PA certificate (Content Credentials). This is essentially the digital proof of origin, which cryptographically proves where an image comes from and how it was edited. And now, hold on tight: Google is a member of the very C2PA initiative that developed this standard and was launched by Adobe, the New York Times, and Twitter. So, they co-developed a standard for authenticity, but conveniently ignore it in their products. Adobe correctly recognizes the image as “human” – the user is now caught between two competing 'truths'.
When photographer Berg dutifully reported this system error to Google, the response came after only 60 seconds, suggesting an automated rejection: “Won't Fix (Intended Behavior)”. In German: “Will not be fixed (intended behavior)”. Google is thus communicating that it is not an error, but intentional, when authentic works are falsely branded as AI products. This is a slap in the face for photographers and journalists. It means that the algorithm defines the truth and there is no right to object. If the digital referee whistles, the game is over – even if he's standing on the wrong field.
MFT Alliance Grows: Sonida from Shenzhen Joins
While a tech giant undermines trust in digital images, there is tangible and more pleasing news on another front. The Micro Four Thirds Alliance (MFT), led by OM Digital Solutions (formerly Olympus) and Panasonic, has a new member: Shenzhen Sonida Digital Technology Co, Ltd. from China. For those who are now raising their eyebrows and asking “Who?”, let it be said: this is potentially very good news for all MFT users.
Sonida is not an unknown entity, but has already produced cameras for other international brands as an ODM manufacturer. So they know how to build cameras. Under their brand “Songdian”, they now want to bring MFT-compatible products to market. For the MFT system, which often competes as an agile underdog against the full-frame dominance of Sony, Canon, and Nikon, this is a win. Every new partner strengthens the ecosystem and promises more choice in cameras and, above all, in lenses. And who knows, perhaps a new player from China will also bring about a refreshing price dynamic. It remains to be seen whether the products will sound as the brand name suggests.
Videos by heise
To Read: When Error Becomes Method
For those who want to delve deeper into the absurd world of AI detection and platform responsibility, Georg Berg's full report on his website “Tellerrand-Stories” is highly recommended. He not only presents the case but also provides the complete technical analysis and the chain of evidence. It is an important document that shows why voluntary self-commitments by tech corporations are often worth no more than the digital paper they are written on. A wake-up call for all who care about authenticity in photography.
(tho)
