惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

The GitHub Blog
The GitHub Blog
S
SegmentFault 最新的问题
MyScale Blog
MyScale Blog
有赞技术团队
有赞技术团队
V
Visual Studio Blog
T
The Blog of Author Tim Ferriss
爱范儿
爱范儿
Vercel News
Vercel News
让小产品的独立变现更简单 - ezindie.com
让小产品的独立变现更简单 - ezindie.com
Y
Y Combinator Blog
Blog — PlanetScale
Blog — PlanetScale
D
DataBreaches.Net
美团技术团队
Microsoft Security Blog
Microsoft Security Blog
大猫的无限游戏
大猫的无限游戏
Cyber Security Advisories - MS-ISAC
Cyber Security Advisories - MS-ISAC
酷 壳 – CoolShell
酷 壳 – CoolShell
GbyAI
GbyAI
A
About on SuperTechFans
云风的 BLOG
云风的 BLOG
The Cloudflare Blog
宝玉的分享
宝玉的分享
V
V2EX
Microsoft Azure Blog
Microsoft Azure Blog

Fastly Blog

Fastly Fastly Fastly Fastly Fastly Fastly Fastly Fastly Fastly Fastly Fastly Fastly Fastly Fastly Fastly Fastly Fastly Fastly Fastly Fastly Fastly Fastly Fastly Fastly Fastly Fastly Fastly Fastly Fastly Fastly
Fastly
Steve Lyons · 2026-08-04 · via Fastly Blog

On August 2, 2026, content provenance stopped being a best practice and became the law. That's the date California's SB 942 (the California AI Transparency Act) became enforceable, with Article 50(2) of the EU AI Act following after a short 3-month grace period. These laws carry real penalties for providers of AI that generates or alters media without provenance metadata identifying the content as AI-generated. While not all organizations are subject to these specific laws, transparency and related regulations are getting a lot of attention and increasing adoption globally. Consumers are also pushing for better transparency. According to a recent report, more than 90% of consumers expect brands to disclose AI usage in marketing. 

This may seem daunting, but there’s good news! Fastly Image Optimizer can help support your ability to preserve and pass through content provenance data through your delivery pipeline. C2PA support for the latest spec (2.4) is now available to all Image Optimizer customers, across all image formats (except JPEG XL), and still at no additional cost.

Why this matters now

Regulators didn't invent these rules in a vacuum. Generative AI made it trivial to produce a photorealistic image in seconds, and just as easy to pass one off as real. When anyone can generate a convincing photo of an event that never happened, "seeing is believing" stops being a safe assumption for your readers, your customers, and your brand. Brand trust is at a premium.

Where C2PA fits

Content provenance methods, such as C2PA, are designed to address this problem. Instead of trying to detect fakes after the fact, C2PA attaches a verifiable record to the file itself: what made it, when, and how it changed along the way. That shifts the burden from "prove this is fake" to "here's exactly where this came from."

The C2PA (Coalition for Content Provenance and Authenticity) is an authority on content provenance, and its metadata spec has effectively become an industry baseline. When people say "provenance metadata," they usually mean C2PA. Fastly and many of our customers are members, because a trustworthy internet is a group project.

Provenance that survives transformation

Here's the hard part most teams run into - with C2PA anything that touches an image, such as resizing, reformatting, or optimizing, can strip or invalidate metadata. That's a problem when your delivery pipeline exists to transform images.

Image Optimizer solves this with C2PA pass-through built directly into the transformation path. Your provenance metadata is preserved through the image optimizations you already run, which means:

  • Provenance that persists through transformations. Resize, reformat, and optimize while the C2PA manifest travels with the image, along with a new attestation added by Image Optimizer.

  • Broad format coverage. It works across the formats you actually use; all image formats (except JPEG XL) supported by Image Optimizer are also supported for C2PA. 

  • No rework or enablement required. Keep your current setup and follow the docs to use pass-through C2PA.

  • Use the tools you already have. Any standard C2PA metadata, up to version 2.4, in your origin files can be passed through Fastly Image Optimizer and re-signed with additional provenance data for the transformations made by our systems. 

Available for all, starting now

Last year, we released C2PA support in Image Optimizer with support for JPEG. What's changed is that C2PA support is now more easily available for everyone, we now support the latest C2PA spec, and we’ve expanded the supported image formats. Now, every Image Optimizer customer can use C2PA without requesting enablement; it's simply part of the product. If you're on IO, you're ready to get started.

How it works

To make images as small as possible, Image Optimizer’s default behavior is to strip the embedded metadata from processed images. The metadata parameter is used to select which type of metadata to pass through the optimization. New c2pa and copyright,c2pa values for the metadata parameter instruct Image Optimizer to preserve the C2PA manifest. 

Keep in mind that the C2PA manifest will result in the optimized image being slightly larger than if the metadata was stripped out. The exact size will depend on the C2PA manifest in the source image and how many attestations it already includes. Image Optimizer’s attestation will add about 12.5 KB plus the size of an embedded thumbnail, depending on the changes that are made to the image.

What's next

This is a foundation, not a finish line. We continuously evolve our products to adapt quickly to regulatory requirements and C2PA standards. We're already exploring external C2PA manifests so you get full provenance without metadata increasing file sizes, and latency. 

Ready to get started?

If you're already using Image Optimizer, C2PA support is available. Check the metadata documentation to put it to work. If you're not, talk to an expert today.