bots don't cry
Autoregressively sewing and generating garments from text guidance
- C2PA has been a success. Its flaws were obvious when the AI industry sold it to governments as a replacement for real regulation. They knew it was not working and lied about it. They avoided regulation. C2PA likely helped. And that's how you make money with C2PA. www.theverge.com/podcast/8740...
- Adobe is so desperate! It is joining the AI cash burning train by subsidizing its entire genAI models wrapper (Firefly). Poor Adobe Stock 🤣 blog.adobe.com/en/publish/2... If you truly think you need genAI (you don’t), there are plenty of better subsidized wrappers slopping around.
- As of February 1, Google has enabled SynthID (invisible watermark layer for genAI) across all Gemini plans/APIs/partners. C2PA remains, with a proprietary tag that Adobe’s verifier can’t surface. Anyone running serious botfarms pipelines knows it’s still removable, at the cost of extra computation.
- Fun fact for anyone who prefers cannibalism over dogfooding for their AI slop needs: The hero video on the infamous "Grok Imagine" landing page was actually generated by Midjourney. cdn.midjourney.com/video/c68d6a...
- Spot the differences between LinkedIn’s updated and previous C2PA AI‑labeling interface! After Meta, and like Adobe, Microslop is probably axing C2PA after 2 years of #abandonware.
- Andrew Jenks appears to be having an existential crisis over C2PA and is now looking for anyone with “concerns” to make him feel less wrong. At this point, legal help might be more useful, considering the possibility he misled both the U.S. government and consumers with vaporware.
- The Cloudflare C2PA service is back online, only took a month. It is non‑compliant, non‑conformant, and "interoperable" exclusively with their friends at Adobe, on a deprecated version. Results for the "chain of provenance" promise for ChatGPT, Windows Paint, Google Gemini and Adobe Firefly:
- The Ring webcam,famous for its security, has rolled out C2PA exactly as expected: non‑compliant and incompatible. Probably for the best, since with server‑side signing it’d take about an hour before YouTube filled up with gen‑AI proudly stamped "Authentic Ring Footage." blog.ring.com/about-ring/y...
- Another @adobe.com sponsored C2PA promo piece, promoted by @andyparsons.net on LinkedIn. And this time @contentauth.bsky.social really cut costs: the "journalist" is literally AI‑generated. If only there were some magical technology to detect AI‑generated images. www.findarticles.com/ring-launche...
- Another @adobe.com sponsored C2PA promo piece, promoted by @andyparsons.net on LinkedIn. And this time @contentauth.bsky.social really cut costs: the "journalist" is literally AI‑generated. If only there were some magical technology to detect AI‑generated images. www.findarticles.com/ring-launche...
- C2PA introduced CAWG in December 2023 to replace the removed TDM assertions in the 2.0 spec, but CAWG is effectively an Adobe feature, created by the same Adobe C2PA team and shipped inside Adobe’s "open‑source" C2PA SDK. Regulators should not be misled: TDM is part of C2PA.
- I wonder why C2PA decided to encode their manifest/signature for plain text in a way that is not visually rendered while they could have used Unicode emojis.🤔 According to C2PA: "the size increase is negligible relative to the asset itself and is a reasonable tradeoff for added security and trust."
- A new study from the Media and Journalism Research Center found that nearly 90% of late‑2024/2025 posts on LinkedIn and Facebook were mostly or fully AI‑generated. While Microsoft and Meta pushed governments to adopt #C2PA to combat AI slop, they were actively welcoming it on their own platforms.
- The Interactive Advertising Bureau (IAB) has rolled out an “AI Transparency and Disclosure Framework” (with #C2PA 😬) to help brands advertising on social platforms where the majority of posts are genAI and a significant share of the user base are automated accounts. www.iab.com/guidelines/a...
- The latest C2PA specifications repurpose portions of the Unicode standard to tag plain text as AI‑generated or modified. The C2PA working group (Adobe, Microsoft, Google, Meta..) never checked whether this would conflict with Unicode’s own guidance, and Unicode is not happy about it.
- The head of Instagram is posting unlabeled AI‑generated text on the very platform flooded with AI slop, while calling for… labeling AI‑generated content. Instagram actually introduced genAI labeling (via C2PA) nearly two years ago, but quietly UI buried it after Adobe’s genAI victims pushed back.
- Most 2024–2025 C2PA pilots have been abandoned, drowned out by the AI slop produced by the C2PA Steering Committee companies. Whatever survives will likely be buried by the conformance program. The CAI Director’s claims are almost comical given Adobe has probably already cut its budget.
- Google’s C2PA AI Slop provenance is bloated beyond the absurd: 2.1 MB of metadata on a 700 kB JPEG, offering nothing useful beyond Google’s own entries and nearly crashing Adobe’s verifier. A simple max 4 kB signed “Generated by AI” tag would’ve been more than enough.
- Ordinarily, a Cloudflare outage has immediate, visible effects across the Internet. Yet its C2PA service has been down for nearly a week with almost no reaction. The reason is simple: outside of government‑driven advocacy, C2PA sees little real‑world use.
- Don’t rely on Adobe Illustrator, like IPTC, for your C2PA manifests, It produces SVG files so bad that they manage to crash Adobe’s own C2PA verifier UI. Use Inkscape, It just work at 1/100 (?) of the file size. IPTC is heavily bankrolled by Adobe, I get they have to use Illustrator.
- The IPTC "Origin Verify" ( @adobe.com C2PA) verifier has returned, promising trust for journalists. Sadly, one of my forgery managed to crash the verifier. But don’t worry, it’s ‘Open Source! Which in @adobe.com novlang means locked down and proprietary.
- To make the LinkedIn’s slop look even sloppier, Microsoft’s PR squad proudly upscales their blurry JPEGs through ChatGPT and then brag about transparency in the name of C2PA.
- Google’s AI image verification is downright terrifying. Shutting it down seems like the only option. "This image is an authentic, historical photograph and is *not AI-generated*." blog.google/technology/a...
- Great! Google rolls out its shiny new AI image verification. So let’s put it to the test with two genAI images: one from DALL·E 3, and one from the latest Nano Banana 3 The verdict? It fails spectacularly. Worse: it ends up granting legitimacy to the very fakes it’s meant to expose.
- It was inevitable. Google made some attempts with the Pixel 10, but Vivo has become the first to certify as 'authentic' an image entirely reconstructed by AI. With that, C2PA can no longer be considered a reliable proof of authenticity.
- Vivo has quietly begun embedding C2PA into photos captured by the camera of its latest flagship, the vivo X300 Pro. However, the C2PAs are not recognized by the official "open-source" online Verifier and c2patool. - both abandontained by @adobe.com @contentauth.bsky.social -
- Vivo has quietly begun embedding C2PA into photos captured by the camera of its latest flagship, the vivo X300 Pro. However, the C2PAs are not recognized by the official "open-source" online Verifier and c2patool. - both abandontained by @adobe.com @contentauth.bsky.social -
- Confirmed! 😀 alphauniverse.com/stories/sony...
