iiinova — where your digital self keeps evolving

bots don't cry @botsdontcry1.bsky.social · 14h

In a cute move. Adobe has removed embedded Content Credentials from all its latest AI generated images and now hides them in its own cloud. It ensures that their AI images won't get flagged on social platforms, while still being "compliant". And Adobe can blame the platforms if anything goes wrong.

000

bots don't cry @botsdontcry1.bsky.social · 21/09/2026

Which image has the right provenance ? 😅 And by the way: "Under Siege, Ukraine's Public Broadcaster Built a Defense Against Deepfakes" is nonsense and smells genAI like it's 2023.

010

bots don't cry @botsdontcry1.bsky.social · 21/09/2026

They did it! The @adobe.com subsidized @pixelstream.bsky.social was on duty. Where are signed photos originating from the Sony cameras @pixelstream.bsky.social Tim? All I see are Photoshop edited images, thanks to exiftool: not c2pa. lies, lies, lies. contentauthenticity.org/blog/under-s...

contentauthenticity.org

Under Siege, Ukraine's Public Broadcaster Built a Defense Against Deepfakes

How Suspilne became one of the first news organizations to bring visible Content Credentials to readers

000

bots don't cry @botsdontcry1.bsky.social · 20/09/2026

Nikon has C2PA support no more, except if you use the "conformant" Adobe @contentauth.bsky.social. A well deserved firmware upgrade for my beloved COOLPIX 100 contentauthenticity.adobe.com/inspect?sour...

$𝗜𝗡 𝗧𝗛𝗘 𝗖𝗢𝗨𝗥𝗧 𝗢𝗙 𝗣𝗨𝗕𝗟𝗜𝗖 𝗢𝗣𝗜𝗡𝗜𝗢𝗡 & 𝗖𝗢𝗥𝗣𝗢𝗥𝗔𝗧𝗘 𝗦𝗔𝗡𝗖𝗧𝗜𝗧𝗬　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　IN RE: THE TOTAL AND ABSOLUTE ABSOLUTION OF THE 𝔽𝕚𝕔𝕕𝕖𝕡™ GENERATIVE ARTIFICIAL INTELLIGENCE MODEL　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　AFFIDAVIT OF INCONTESTABLE ETHICS AND FORENSIC LEGAL BLANKETING　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　𝗜, 𝗖𝗼𝘂𝗻𝘀𝗲𝗹 𝗳𝗿𝗼𝗺 𝘁𝗵𝗲 𝗖𝗼𝗿𝗽𝗼𝗿𝗮𝘁𝗲 𝗖𝗼𝘂𝗻𝘀𝗲𝗹 𝗳𝗼𝗿 𝗣𝗿𝗼𝗳𝗶𝘁 𝗔𝗰𝗰𝗲𝗹𝗲𝗿𝗮𝘁𝗶𝗼𝗻 (C2PA), acting on behalf of Adobe’s Department of Immutable Compliance and Copyright Shielding, do hereby attest, swear, and state under penalty of severe public relations fallout, the following truth-adjacent facts:　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　𝗪𝗛𝗘𝗥𝗘𝗔𝗦, the generative artificial intelligence model known commercially as 𝔽𝕚𝕔𝕕𝕖𝕡™ has been trained with a level of ethics so pure, so pristine, and so completely beyond reproach that it borders on the celestial, having absorbed only datasets that were voluntarily surrendered, legally loopholed, or otherwise rendered compliant via 47 pages of unread Terms of Service Agreements;　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　𝗪𝗛𝗘𝗥𝗘𝗔𝗦, the proprietary training architecture of 𝔽𝕚𝕔𝕕𝕖𝕡™ has been thoroughly scrutinized, algorithmically laundered, and formally blessed by a highly compensated tribunal of Adobe Lawyers, whose vision is based entirely on billable hours and whose definition of \"fair use\" is whatever we can successfully defend in a federal circuit court;　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　𝗪𝗛𝗘𝗥𝗘𝗔𝗦, any resemblance between the outputs of 𝔽𝕚𝕔𝕕𝕖𝕡™ and the copyrighted, hard-fought, uncompensated works of living human creators is purely coincidental, statistically inevitable, and, frankly, a massive compliment to the original artists;　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　𝗧𝗛𝗘𝗥𝗘𝗙𝗢𝗥𝗘, 𝗕𝗘 𝗜𝗧 𝗥𝗘𝗦𝗢𝗟𝗩𝗘𝗗, that 𝔽𝕚𝕔𝕕𝕖𝕡™ is hereby declared legal…$

000

bots don't cry @botsdontcry1.bsky.social · 18/09/2026

Playing around with Adobe @contentauth.bsky.social latest update of their "verifier" with my beloved Leica SL3-P. Adobe Content Authenticity "Inspect" is now the official C2PA verifier: no one maintain the open-source one. contentauthenticity.adobe.com/inspect?sour...

No Leica Camera was harmed, it's just the Adobe verifier getting vibe coded.

001

bots don't cry @botsdontcry1.bsky.social · 18/09/2026

Beh, it looks like the post has been scrapped, probably no one wanted to take responsibility for it, or they just run out of tokens.🤔

000

bots don't cry @botsdontcry1.bsky.social · 18/09/2026

Can't wait for Adobe @contentauth.bsky.social 's post on successful C2PA use in its marketing software. As predicted, mass surveillance & hyper-personalization can trace the origin and distribution path of the ads, C2PA verified experienceleague.adobe.com/en/docs/gens... bsky.app/profile/bots...

experienceleague.adobe.com

Content Credentials for Organizations | Adobe GenStudio for Performance Marketing

Learn about how to apply and review Content Credentials in GenStudio for Performance Marketing.

022

bots don't cry @botsdontcry1.bsky.social · 16/09/2026

Not bragging, but I probably have Apple staff in my long list of followers 😅 The harsher reality is that Apple is not in the generative AI nonsense, like all the other ones promoting C2PA. And they probably had leverage to get exclusivity for the Sony 🔏 sensor. security.apple.com/blog/apple-r...

This is not a simple problem to address. Modern cameras rely on sophisticated image-processing algorithms to produce the final viewable image, so certifying that an image accurately reflects what a real camera sensor captured requires a chain of trust covering the sensor as well as the computational photography software that interpreted the capture. Industry approaches to this problem, based on the C2PA standard, attach provenance metadata after capture and certify the history of image edits from that point forward. This approach, however, is vulnerable to compromise at any point in the editing chain, and a viewer has no way to detect such a failure. It can also create privacy risks for photographers working in dangerous conditions by tying the image to a public identity, either to a particular device or to an individual.

000

bots don't cry @botsdontcry1.bsky.social · 11/09/2026

Can't wait for @contentauth.bsky.social 's post on successful C2PA use in a war zone. As predicted, suspilne.media is now publishing C2PA-signed GPS coordinates of their journalists pictures. fotoforensics.com/analysis.php... fotoforensics.com/analysis.php...

000

bots don't cry @botsdontcry1.bsky.social · 07/09/2026

The @iptc.org ’s "Origin Verify" tool is messing up its C2PA verifier again, trusting a random self-signed certificate and giving it legitimacy 😂. For many reasons, this is starting to look like sabotage 💰. It's 2026 and there is still not a single working C2PA verifier. bsky.app/profile/bots...

This is the 3rd of 12 ingredients in the media asset's lifecycle, identified by the unique ID a5b2c2d8-22a8-416c-bc08-d11f05222e53.
The content was created using xAI Grok Imagine (specifically Grok Imagine version 0.0.0)
The asset was signed using a Self-signed ephemeral certificate provided via the Content Authenticity SDK.Trust Limitation: The signature explicitly carries a "LOCAL USE ONLY" flag, meaning the self-signed certificate is meant for development or local testing and will not be recognized as a trusted authority by public C2PA validators.

001

bots don't cry @botsdontcry1.bsky.social · 05/09/2026

The Google C2PA Core Group Product Manager @sherifhanna.bsky.social has tasked an AI to illustrate the release of a C2PA C++ Library. The result ? An octagon, representing absolute spatial control, total surveillance, and the rigid imposition of state order. developers.googleblog.com/introducing-...

010

bots don't cry @botsdontcry1.bsky.social · 02/09/2026

Anthropic has signed the EU AI Act's Article 50(2) Code of Practice on Transparency of AI-Generated Content: Good boys. They use C2PA for signed provenance metadata. The metadata completely forgets to actually label the content as AI-generated. It is not interoperable, as always with C2PA. Done.

{
"action": "com.anthropic.claude.provided",
"softwareAgent": {
"name": "Claude"
},
"parameters": {
"com.anthropic.origin-confidence": "unknown"
},
"description": "Claude provided this file at the request of a user and may have created or modified the file contents."
}

000

bots don't cry @botsdontcry1.bsky.social · 05/08/2026

"However, the intention to enforce so-called content credentials as an attribute of credibility also reveals the well-known walled garden mentality of market-dominating technology companies." by @evisualyn.bsky.social with a nice electronic waste part. ahnp.ub.uni-heidelberg.de/journals/dah...

ahnp.ub.uni-heidelberg.de

A Brief History of Authenticity Infrastructures, 1994 to 2024 | International Journal for Digital Art History

021

bots don't cry @botsdontcry1.bsky.social · 03/08/2026

The Adobe’s Content Authenticity bros have updated their verifier UX, after 6 years... It looks like they have expanded the attack surface... My 2 years old forgeries are working again! 🥳 Can you guess which Content Credential is forged, just by looking at the left column? 😬

100

bots don't cry @botsdontcry1.bsky.social · 03/08/2026

The Adobe’s C2PA bros have published a guide on using Content Credentials under new AI regulations. They managed to break it immediately by throwing in a custom `asset-type-choice` rule for inference parameters.🍹 No compliant verifiers will actually be able to read it. c2pa.org/a-new-implem...

000

bots don't cry @botsdontcry1.bsky.social · 01/08/2026

Higgsfield AI, another genAI wrapper, has started to use C2PA for its slop. It's not interoperable and supports only the legacy trust list. In short: Nothing for the end-user the C2PA is aimed at. But C2PA has always been for regulators: Higgsfield AI will probably be compliant with the EU-AI Act.

Nano Banana wrapped into Higgsfield AI C2PA. Somehow the "verifier" decided that the file was tampered, while it's not: It's only Higgsfield AI lacking the Google C2PA trustlist.

Seedream 5.0 pro C2PA, wrapped with Higgsfield AI C2PA.
Somehow the "verifier" decided that BytePlus was on a trust list... while it's not.

010

bots don't cry @botsdontcry1.bsky.social · 17/07/2026

Please meet @andyparsons.net , Adobe Executive who risks journalists' lives to pump his stock portfolio. He said to @marinaadami.bsky.social: "With the exclusion of GPS [...], the Google Pixel phone will sign all [...]" That's a lie: proof in his own repo: raw.githubusercontent.com/contentauth/...

100

bots don't cry @botsdontcry1.bsky.social · 16/07/2026

3 multi-billion dollars AI companies. 2 Certificate Authorities. All bound by Google’s C2PA conformance program. A single chain of trust. A mille-feuille of failures. Image 1: Open AI genAI, SSL.com certificates Image 2: Open AI genAI, Trufo certificates

001

bots don't cry @botsdontcry1.bsky.social · 15/07/2026

Is the BBC being weaponized as a salesman for the AI industry? Despite years of C2PA failures, the BBC continues to push Adobe's narrative, the misleading "Nutrition label", and is now funding this exact, broken setup for Ukraine’s public media during an active information war. 🤮

000

bots don't cry @botsdontcry1.bsky.social · 15/07/2026

The BBC press release! Claiming "an unbroken chain of trust from the first shot through to publication". It requires all parties compatible trust lists, and implementations. Sony 📸certs are for Sony Ci Media Cloud workflows only💰, can't be used with the Adobe Cloud. www.bbc.co.uk/mediacentre/...

bbc.co.uk

Providing Ukrainian audiences with content they can trust

A new ‘nutrition label’ for its content, ensuring audiences receive information they can trust in wartime

100

bots don't cry @botsdontcry1.bsky.social · 11/07/2026

I found the demo: the image is not trusted by the official verifier, certificates missing. They patched it on their demo with their own certificates. With a nice biased BBC survey, to claim later that C2PA increase trust. Image taken 2026-05-31 + Photoshop & Lightroom suspilne.media/1333984-pere...

Exiftool:
Make : SONY
Camera Model Name : ILCE-1M2
Orientation : Horizontal (normal)
Samples Per Pixel : 3
X Resolution : 300
Y Resolution : 300
Resolution Unit : inches
Software : Adobe Photoshop 26.10 (Macintosh)
Modify Date : 2026:06:18 13:07:29
Artist : news
Copyright : suspilne
Exposure Time : 1/1000
F Number : 2.8
Exposure Program : Manual
ISO : 100
Sensitivity Type : Recommended Exposure Index
Recommended Exposure Index : 100
Exif Version : 0232
Date/Time Original : 2026:05:31 16:08:13
Create Date : 2026:05:31 16:08:13
Offset Time : +03:00
Offset Time Original : +03:00
Offset Time Digitized : +03:00
Shutter Speed Value : 1/1000
Aperture Value : 2.8
History Software Agent : Adobe Photoshop Lightroom Classic 14.5.1 (Macintosh), Adobe Photoshop 26.10 (Macintosh)
Adobe Camera Raw:
Crop Angle -0.4
Exposure -0.15
Contrast 5
Highlights -20
Shadows 22
Whites 15
Blacks -5
Vibrance 5
Clarity 5
Dehaze 5
Sharpening 40
Noise Reduction 15
Color Noise Reduction 25
Enable Lens Corrections On

000

bots don't cry @botsdontcry1.bsky.social · 10/07/2026

Looks like the BBC is spinning their C2PA implementation disaster. And @contentauth.bsky.social is in need of PR since they can't afford Wired advertorials anymore. It’s not working! 😬 but a total win for the journalists who got 2 new cameras out of the deal. corp.suspilne.media/en/top-stori...

Yes, it's a bit expensive for 2 cameras and 2 lines of javascript, I hope they could grab as much as lenses as possible and were not forced to buy Adobe or Sony subscriptions

100

bots don't cry @botsdontcry1.bsky.social · 10/07/2026

Bing Image Creator has lost its Content Credentials, no more C2PA for its users. A moment of silence for Microsoft Responsible AI Governance Framework, just in time to troll the EU AI Act. Why retiring a flawed technology?, couldn't Copilot fix it?

000

bots don't cry @botsdontcry1.bsky.social · 08/07/2026

I have updated my corrections on page 9, as it looks you can be on the official "conforming" product list without a product. It's vaporware. In the case of Xiaomi and Qualcomm, they don't have any trusted certificates. Why would you do that? 😂

000

bots don't cry @botsdontcry1.bsky.social · 08/07/2026

The "scenarios" depicted in the report are already the reality of C2PA, conformant or not. page 9 about "conforming product" is really misleading, up to the url. The conformance list in short: Google. An advertising company, for a report on c2pa privacy. 😅 library.witness.org/product/c2pa...

000

bots don't cry @botsdontcry1.bsky.social · 08/07/2026

New C2PA report! The authors probably had to walk on eggs to please the @contentauth.bsky.social plutocracy. But I can feel the satire: "The value of the ecosystem depends on it being governed as seriously as it is built." 🤣 bsky.app/profile/ingo...

000

bots don't cry @botsdontcry1.bsky.social · 06/07/2026

Of course, The C2PA "Conformant" signers from Google & OpenAI completely wipe out Adobe's provenance, crafted with love & greed, to overwrite it with their own: tracking for Google and nothing for ChatGPT (whose timestamp certificate isn't even trusted to show a creation date).

Google Gemini: add the goose into the design of the first image, don't forget to multiply the eggs

OpenAI ChatGPT: add the goose into the design of the first image, don't forget to multiply the eggs

"informational": [
{
"code": "timeStamp.untrusted",
"url": "self#jumbf=/c2pa/urn:c2pa:828c1382-57e9-4aa5-ac3b-5c9f0c2e602b/c2pa.signature",
"explanation": "timestamp cert untrusted: OpenAI TSA Leaf"
}
],

000

bots don't cry @botsdontcry1.bsky.social · 06/07/2026

Adobe Firefly C2PA update! Adobe has brought back its fancy AI reference thumbnails, probably to impress some regulator. In their great kindness, they bundle them as PNG and duplicate them like a goose, probably to impress some trader. That's 7.3MB of AI slop that should only take up 1.0MB.

000

bots don't cry @botsdontcry1.bsky.social · 27/06/2026

󠄳󠄢󠅀󠄱󠅄󠅈󠅄︀︁︀︀︀︄󠇎The latest Leica SL3-P camera ships, of course, with C2PA to prove "authenticity". Except It uses a vintage C2PA: only "interoperable" with other legacy software, like the Adobe ones. But it still does an excellent job proving your Affluentity.

Legacy. Fun fact, you can still edit the exif:DateTime and the xmp:Rating without breaking the signature

000

bots don't cry @botsdontcry1.bsky.social · 25/06/2026

You may wonder how can Adobe @contentauth.bsky.social "Adobe content Authenticity" validates what other C2PA verifiers reject? It should be be the source of truth since it is the only one "conformant"? 🤣 It turns out that "Adobe content Authenticity" validates shit, literally, example:

remote manifest forged, urls can be renamed and redirected at will

000

bots don't cry @botsdontcry1.bsky.social · 24/06/2026

C2PA verifiers in June 2026, in 4 screenshots: ❌ Official one: dead ❌ The “files never leave your device* (*except to call Adobe)”, by @andyparsons.net. ❌ A popular SEO‑bait edition. ✅ The “we upload everything to our servers” Adobe‑conformant. The signed image: shorturl.at/8IgkM Links in ALT

000

bots don't cry @botsdontcry1.bsky.social · 12/06/2026

🎉 Positive C2PA news! - Conformant C2PAs are interoperable with Google Nano Banana (note: PII is not redacted) - A better C2PA verifier by IPTC: originverify.iptc.org (the official Adobe one is misleading & not maintained) - A proper Camera app by @nathan.freitas.net proofmode.org/blog/proofmo... 👏

IPTC News Provenance Verification Origin Verify https://originverify.iptc.org/ displays proper provenance chain and gives access to raw C2PA manifest json data. It shows a Proofmode Camera image being modified by "Google Media Processing Services"

000

bots don't cry @botsdontcry1.bsky.social · 10/06/2026

Adobe is again misleading regulators while promoting the CREATOR Act, claiming C2PA can signal artists’ TDM preferences (Text/Training and Data Mining). blog.adobe.com/en/publish/2... C2PA explicitly clarified to an EU consultation that it cannot. @repyvetteclarke.bsky.social @foushee.house.gov ?

000

bots don't cry @botsdontcry1.bsky.social · 09/06/2026

It took 5 months for @adobe.com @contentauth.bsky.social to take action on a severity "High" "Uncontrolled Resource Consumption vulnerability." affecting their C2PA verifier. The "attacker" was Google Gemini bloated C2PA. Because, you know, C2PA is "interoperable". nvd.nist.gov/vuln/detail/...

nvd.nist.gov

NVD - cve-2026-34665

011

bots don't cry @botsdontcry1.bsky.social · 28/05/2026

The Google C2PA lead is casually mansplaining "How the [Gemini/C2PA] architecture works". If the LLM is trained on conform C2PA for genAI: there was only the Google ones available (filled with proprietary data) at the time of training. This means the C2PA CAN NOT get explained to the end user.

That's not how the architecture works. Under the hood, full C2PA validation is performed by a conformant validator. Its results are then turned over to the LLM for explanation to the end user.

100

bots don't cry @botsdontcry1.bsky.social · 20/05/2026

Even experienced reporters get caught citing AI slop. A @theverge article by @zombiewretch.bsky.social is citing an SEO AI Slop Farm, c2pa(dot)ai as evidence for C2PA adoption. Is there a way to fix this @reckless.bsky.social, @davidpierce.xyz ? www.theverge.com/ai-artificia...

010

bots don't cry @botsdontcry1.bsky.social · 20/05/2026

R.I.P. @adobe.com @contentauth.bsky.social "Adobe Content Authenticity". Deceit was its only consistency, hidden under perpetual betas and manufactured goodwill. "Adobe Content Authenticity" was the final pathetic spasm of a company rotting from the inside. 2024-10-08 - 2026-05-19

000

bots don't cry @botsdontcry1.bsky.social · 19/05/2026

Great! Google has upgraded its AI image verification. So let's try it again with a good old DALL·E 3 image. The verdict? It STILL fails spectacularly. Worse: it ends up granting legitimacy to the very fakes it’s meant to expose.

If Google says so...
No, this image was not generated by AI.

It is a highly famous, authentic historical photograph known as "Tank Man," taken on June 5, 1989, by Associated Press photojournalist Jeff Widener.

The photograph captures a lone protester standing defiantly in front of a column of Type 59 tanks on Chang'an Avenue in Beijing, the day after the Chinese military's violent suppression of the Tiananmen Square protests. It remains one of the most iconic images of resistance and courage in modern history.

181

bots don't cry @botsdontcry1.bsky.social · 19/05/2026

Great! OpenAI has rolled out a C2PA verifier, "for safer, more transparent AI‑slop provenance." The tool, vibe coded under an intern’s armpit, labels any C2PA‑signed AI output as if it were generated with OpenAI’s own systems. The genAI laundering season has officially begun. openai.com/verify/

000

bots don't cry @botsdontcry1.bsky.social · 27/04/2026

The OpenAI C2PA implementation has been totally dysfunctional for more than a month. The new implementation by Trufo AI (with watermarks?) for gpt-image 2 is not any better than the previous @truepic.bsky.social disaster. As always, C2PA is just there to please regulators.

000

bots don't cry @botsdontcry1.bsky.social · 27/04/2026

A research paper about Security Analysis of C2PA and its Implementation, not AI industry sponsored. "Our results show that the specifications and the current implemented C2PA ecosystem do not yet provide the guarantees required for reliable deployment or standards adoption." eprint.iacr.org/2026/804

eprint.iacr.org

Verifying Provenance of Digital Media: Security Analysis of C2PA and its Implementation

Generative AI and advanced editing tools enable malicious actors to create high-quality fake images that can facilitate fraud, attack reputations, and manipulate elections. We analyze security proper-...

100

bots don't cry @botsdontcry1.bsky.social · 22/04/2026

The #C2PA video player that "won" the NAB award is a refactor of original code by Trufo AI that was presented by what's left of BBC & ITN at the #IBC Show 2025. How it ended-up @ebu.ch and CBC/Radio-Canada?🤔 github.com/trufo-ai/c2p... github.com/Security4Med... tmbroadcast.com/ebu-cbc-radi...

100

bots don't cry @botsdontcry1.bsky.social · 15/04/2026

It's an Adobe promo. I’m not sure what the point of the article is, aside boosting other Adobe researches and bringing up a 2 year old "Durable Content Credentials" solution that Adobe tried, implemented alone, and then dropped because it failed. 🤷 A least better than advertising on content farms.

000

bots don't cry @botsdontcry1.bsky.social · 11/04/2026

Would you put your trust in an immature solution from the AI industry, the one being promoted here by @contentauth.bsky.social advisor Henry Ajder, or in ExifTool, a battle‑tested tool that local journalists have relied on for decades, assuming the same AI industry hadn’t wiped out their revenue?

The story being told that people doubted the image because "AI has eroded trust" is a convenient narrative and does not change the fact that the metadata is authentic and externally verifiable.

You don't have to rely on speculative "AI distrust" narratives.

AI tools can spoof a few EXIF fields, but they cannot recreate the full, coherent set of hardware‑level fields that match a real camera’s imaging pipeline.

The metadata is from SWNS, a commercial media agency: but still verifiable.
any professional journalist, photo editor, or newsroom can:
Request the original file
Inspect the EXIF/IPTC metadata
Verify the Pixel 6 motion‑photo structure
Confirm the timestamps and workflow history

020

bots don't cry @botsdontcry1.bsky.social · 01/04/2026

Meet JR Delaney, the "journalist" that C2PA evangelists love to cite. Why? He’s an LLM trained on Adobe press releases and SEO. Obviously A stands for AI in C2PA, not authenticity.

000

bots don't cry @botsdontcry1.bsky.social · 31/03/2026

191 bytes VS 59.6kB, C2PA needs 300X more data for text. It already breaks the open-web, your favorite data-center next?

000

bots don't cry @botsdontcry1.bsky.social · 31/03/2026

Thanks to EU regulators, the latest draft of the AI Act drops the C2PA reference. But Erik Svilitch double down and submitted a public comment, full of nonsense. The fun part: You can't copy‑paste URLs from his C2PA‑signed PDF: his Unicode hack breaks the open web. encypher.com/tools/verify...

000

bots don't cry @botsdontcry1.bsky.social · 30/03/2026

"[...] we show how provenance and authenticity standards are leveraged to both allow big tech to demonstrate commitments to “responsible AI” to preempt regulatory intervention and to build an infrastructure through which their preexisting dominance is reinforced. "🍿 nomadit.co.uk/conference/e...

nomadit.co.uk

Critical AI and infrastructural governance: C2PA and the construction of global provenance and authenticity standards

The paper analyses the efforts of the Coalition for Content Provenance and Authenticity (a initiative for provenance and authenticity standards), to identify a

010

bots don't cry @botsdontcry1.bsky.social · 30/03/2026

Overlooked in this 🔥 article by @zombiewretch.bsky.social www.theverge.com/ai-artificia...: The adverting Agency Publicis! Publicis, a proud steering Committee member of the C2PA, has a long history of working with both Samsung and Tiktok. ads.tiktok.com/business/en/...

It’s important to note that both Samsung and TikTok are members of the Content Authenticity Initiative, a group that aims to make content authenticity and transparency “scalable and accessible” by promoting the industry-wide adoption of C2PA. That means TikTok and Samsung supposedly share similar ideals regarding the labelling of AI content. If Samsung knowingly used AI to make its videos, it should have told TikTok when the ads were submitted. If TikTok was informed, it should have made sure its users were aware, per the platform’s own advertising policies.

100