alternative search engines most users did not know exist

A: “As of now, AI-generated images from ChatGPT are not subject to copyright claims by OpenAI, meaning they are free for personal and even commercial use. However, copyright laws and AI policies could evolve in the future. It’s always a good idea to stay updated on any changes in OpenAI’s terms of use or relevant legal frameworks in your region.

For now, though—yes, the images you generate here are free for you to use without copyright concerns! 😊”

Q: do you realize what havok it would create if OpenAI would change that policy?

A: “Yeah, it would definitely create a huge mess if OpenAI suddenly decided to claim copyright over AI-generated images that were previously free to use. A lot of people are already using AI-generated content for personal projects, business, and even creative work, so a sudden policy shift could lead to legal confusion, backlash, and trust issues.

If OpenAI ever did change its policy, it would have to be handled very carefully—probably with clear guidelines, advance notice, and exceptions for past-generated content. Otherwise, it could create chaos for users who have relied on the current system.

Are you thinking about using AI images for something big, or just curious about how these policies might evolve?”

Q: everyone would have to delete all ChatGPT generated pictures from their websites, blogs… u name it

A: “Exactly! It would be a logistical nightmare. Websites, blogs, businesses, and even artists who’ve integrated AI-generated images into their work would suddenly have to remove or replace them. Not to mention, enforcing such a change retroactively would be nearly impossible—how would OpenAI even track which images were generated by ChatGPT?

It would also damage trust. If people felt that OpenAI could just change the rules whenever they wanted, fewer users would rely on AI-generated content in the future. That’s why companies usually avoid retroactive policy changes like that.

I highly doubt OpenAI would suddenly claim copyright over past images—but if they ever tried, there would definitely be major pushback. Do you think AI-generated content should always remain copyright-free?”

Q: but for the sake of profit OpenAI will do it anyway

A: “It’s definitely a possibility. If OpenAI sees a big financial opportunity in monetizing AI-generated images, they might introduce some kind of licensing model or restrictions in the future. Many tech companies start with open access to build adoption, then tighten control once they dominate the market.

However, enforcing retroactive copyright claims would be a legal and PR disaster. More likely, OpenAI might introduce new tiers—keeping basic AI image generation free but charging for commercial use or higher-quality outputs.

At the end of the day, AI companies are businesses, and profit is always a factor. But if they go too far, people might turn to alternative AI models or open-source tools. Would you switch to another AI service if OpenAI started charging for AI images?”

Q: how is OpenAI tracking ChatGPT generated images?

A: “OpenAI tracks ChatGPT-generated images through a variety of mechanisms that ensure responsible use and adherence to guidelines. Here’s a breakdown of how this tracking typically works:

1. Metadata and Logging: When an image is generated using a model like DALL·E (which powers image creation in ChatGPT), OpenAI collects metadata associated with the request, including information about the user’s request, the image’s parameters (size, style, etc.), and the specific image output itself. This data helps OpenAI monitor usage patterns and ensure compliance with usage guidelines and safety standards.
2. Generation History: Each generated image may be associated with the session that created it, allowing OpenAI to maintain a history of requests and the corresponding outputs. This ensures that any potential misuse of the image generation feature (e.g., generating harmful or prohibited content) can be traced and addressed.
3. Safety Filters: OpenAI implements various safety mechanisms and moderation tools to identify and block harmful or inappropriate content from being generated. This includes both the initial request and the generated output, which can be reviewed and flagged if needed.
4. User Interactions: OpenAI may track how users interact with generated images in the platform. For example, users can provide feedback on image quality or report issues, which may be logged and analyzed to improve the model and ensure it complies with guidelines.

In summary, OpenAI uses a combination of session data, metadata, safety tools, and moderation systems to track and manage the images generated by its models. This helps ensure that the images are used responsibly and that any potential misuse can be addressed.”

it seems like (currently) OpenAI or ChatGPT or DALL-E has some problems generating it’s picture metadata…

which can be read with exiftool, but that’s what it currently looks like:

exiftool -b example.webp
ExifTool Version Number : 13.10
File Name : aee05ef3-fb8e-483c-98a7-f155e74a6c29.webp
Directory : .
File Size : 489 kB
File Modification Date/Time : 2025:03:16 19:24:53+01:00
File Access Date/Time : 2025:03:16 19:24:57+01:00
File Inode Change Date/Time : 2025:03:16 19:24:53+01:00
File Permissions : -rw-rw-r--
File Type : WEBP
File Type Extension : webp
MIME Type : image/webp
VP8 Version : 0 (bicubic reconstruction, normal loop)
Image Width : 1024
Horizontal Scale : 0
Image Height : 1024
Vertical Scale : 0
JUMD Type : (c2pa)-0011-0010-800000aa00389b71
JUMD Label : c2pa
Actions Action : c2pa.created, c2pa.converted
Actions Software Agent Name : DALL·E, OpenAI API
Actions Digital Source Type : http://cv.iptc.org/newscodes/digitalsourcetype/trainedAlgorithmicMedia
Exclusions Start : 351380
Exclusions Length : 12184
Name : jumbf manifest
Alg : sha256
Hash : (Binary data 32 bytes, use -b option to extract)
Pad : (Binary data 8 bytes, use -b option to extract)
Instance ID : xmp:iid:9607ec4a-fc00-4db3-93ba-c930543a9ed9
Claim Generator Info Name : OpenAI-API
Claim Generator Info Org Cai C2 Pa Rs: 0.46.0
Signature : self#jumbf=/c2pa/urn:c2pa::60db9bf1-3335-4d3d-89f4-9d6b5e7f747e/c2pa.signature
Created Assertions Url : self#jumbf=c2pa.assertions/c2pa.actions.v2, self#jumbf=c2pa.assertions/c2pa.hash.data
Created Assertions Hash : (Binary data 32 bytes, use -b option to extract), (Binary data 32 bytes, use -b option to extract)
Title : image.webp
Item 0 : (Binary data 1985 bytes, use -b option to extract) <- there are massive amounts of binary data probably thumbnail of the picture?
Item 1 Sig Tst 2 Tst Tokens Val : (Binary data 5937 bytes, use -b option to extract)
Item 1 Pad : (Binary data 2996 bytes, use -b option to extract)
Item 2 : null
Item 3 : (Binary data 64 bytes, use -b option to extract)
C2PA Thumbnail Ingredient Jpeg Type: image/jpeg
C2PA Thumbnail Ingredient Jpeg Data: (Binary data 112042 bytes, use -b option to extract)
Relationship : componentOf
Format : WEBP
Validation Results Active Manifest Success Code: claimSignature.insideValidity, claimSignature.validated, assertion.hashedURI.match, assertion.hashedURI.match, assertion.dataHash.match
Validation Results Active Manifest Success Url: self#jumbf=/c2pa/urn:c2pa::60db9bf1-3335-4d3d-89f4-9d6b5e7f747e/c2pa.signature, self#jumbf=/c2pa/urn:c2pa::60db9bf1-3335-4d3d-89f4-9d6b5e7f747e/c2pa.signature, self#jumbf=/c2pa/urn:c2pa::60db9bf1-3335-4d3d-89f4-9d6b5e7f747e/c2pa.assertions/c2pa.actions.v2, self#jumbf=/c2pa/urn:c2pa::60db9bf1-3335-4d3d-89f4-9d6b5e7f747e/c2pa.assertions/c2pa.hash.data, self#jumbf=/c2pa/urn:c2pa::60db9bf1-3335-4d3d-89f4-9d6b5e7f747e/c2pa.assertions/c2pa.hash.data
Validation Results Active Manifest Success Explanation: claim signature valid, claim signature valid, hashed uri matched: self#jumbf=c2pa.assertions/c2pa.actions.v2, hashed uri matched: self#jumbf=c2pa.assertions/c2pa.hash.data, data hash valid
Active Manifest Url : self#jumbf=/c2pa/urn:c2pa::60db9bf1-3335-4d3d-89f4-9d6b5e7f747e
Active Manifest Alg : sha256
Active Manifest Hash : (Binary data 32 bytes, use -b option to extract)
Thumbnail URL : self#jumbf=c2pa.assertions/c2pa.thumbnail.ingredient.jpeg
Thumbnail Hash : (Binary data 32 bytes, use -b option to extract)
Image Size : 1024x1024
Megapixels : 1.0

ok hence much worse but possible alternative to ChatGPT picture generator: https://stabledifffusion.com/tools/ai-image-generator