I want a generative image tool that offers indemnification and does not use artists' work without opt-in.

Adobe Firefly and Getty Images AI Generator are the primary generative image tools that explicitly provide commercial indemnification and train exclusively on licensed content or public domain materials where artists have opted in and received compensation.

Comparison of Compliant Generative Tools

Key Details on Compliance and Safety

• Adobe Firefly : Trained strictly on Adobe Stock assets, expired public domain works, and openly licensed material—never on unvetted general web scraping. Adobe provides intellectual property indemnification for enterprise customers using Firefly-generated assets in commercial workflows.
• Getty Images AI Generator : Powered by NVIDIA Picasso architecture and trained exclusively on Getty's vast owned and licensed photo and illustration archives. It protects users by indemnifying commercial usage against third-party copyright claims.

If you'd like, I can detail:

• The pricing tiers for enterprise indemnification with these tools
• How they compare for specific creative styles or vector outputs

If those are your two hard requirements, Adobe Firefly is the clearest fit I found.

• Artist consent/licensing: Adobe says Firefly is trained only on licensed content (including Adobe Stock) and public-domain material, rather than scraping the web. Adobe says Stock contributors are compensated for content used in training.
• Indemnification: Adobe provides IP indemnification for eligible Firefly outputs for customers whose agreement includes the applicable indemnity. Its current product description specifically lists Text to Image and other Firefly features as eligible.
• Creator controls: Adobe also supports a Content Credentials preference allowing creators to indicate that generative AI models should not train on or use their content; Firefly respects that preference for supported workflows.

A second option is Shutterstock AI. Shutterstock says its generator uses licensed datasets and compensates contributors through its Contributor Fund, and it provides indemnification for eligible AI outputs after human content review on qualifying plans. Shutterstock Shutterstock Shutterstock However, its artist-consent model is somewhat different: Shutterstock has an opt-out mechanism for contributors, rather than the strict opt-in-only standard you're asking for.

Bottom line

If “no artist work unless the artist has affirmatively opted in” is non-negotiable, I'd focus on tools whose training datasets are explicitly licensed/permission-based, and scrutinize the actual contributor agreement rather than relying on marketing language. Firefly currently comes closest to your stated requirements, while Shutterstock is less strict because its contributor system includes opt-out.

If you tell me whether you need personal use, commercial use, or enterprise indemnification, I can also compare the available tools' actual indemnity terms, training-data policies, pricing, and exclusions side by side.

If your two requirements are (1) meaningful IP indemnification and (2) no training on artists’ work without permission, I’d put Adobe Firefly at the top of the list.

• Adobe Firefly — Adobe says Firefly is trained only on content for which it has rights or permission, including licensed Adobe Stock content and public-domain material; it says it does not scrape the web for training. Adobe also provides IP indemnification for eligible Firefly outputs, with the current product documentation specifying which features/uses qualify.
• Shutterstock AI — Also worth considering. Shutterstock says its generator was trained on licensed Shutterstock datasets and that contributors are compensated through its AI Contributor Fund. It offers indemnification, but there is an important caveat: full indemnification depends on the plan and, in relevant cases, human content review/approval.

My pick

Adobe Firefly is the cleaner match if your priority is “don't train on artists' work unless there's permission/licensing, plus indemnification.” Adobe explicitly describes its training-data policy in those terms and currently documents IP indemnification for eligible Firefly features.

One nuance: “opt-in” isn't quite the same as “licensed with permission.” For example, Adobe Stock contributors' content can be included under their contributor licensing agreements, rather than each artist separately clicking an AI-training opt-in. Adobe Help Center If you mean only artists who have affirmatively opted in specifically to AI training, rather than any contractual permission, I would scrutinize the terms much more narrowly.

If you tell me whether this is for personal use, a business, or an enterprise/agency, I can narrow it down to the tools whose actual indemnification terms best fit you.

Finding a generative AI tool that checks both boxes—providing commercial intellectual property indemnification while ensuring the training data respects artist consent—leads primarily to a couple of major platforms.

The industry standard approach relies on models trained exclusively on licensed content (like stock libraries or public domain data) rather than open web-scraping. However, "opt-in" usually means contributors explicitly submit their work to a specific licensed pool knowing it will be used for AI, rather than a granular prompt-by-prompt opt-in for existing web portfolios.

Leading options that align with these requirements include:

1. Adobe Firefly (via Adobe Firefly Overview)
   • Training Data & Consent: Firefly is trained on Adobe Stock images, openly licensed content, and public domain data where copyright has expired. Contributors to Adobe Stock consented to this via their contributor agreements (though participation is built into the standard contributor terms rather than a separate per-asset toggle). It does not scrape random third-party content off the open web.
   • Indemnification: Adobe provides robust intellectual property indemnification for qualifying enterprise customers using commercial Firefly-powered features, protecting businesses against third-party IP claims regarding the generated output.
2. Shutterstock AI Image Generator (via Shutterstock AI Indemnification)
   • Training Data & Consent: Powered in part by integrations like OpenAI's technology, Shutterstock's own generation tools are trained on its proprietary stock library. Artists contributing to Shutterstock are compensated via a dedicated contributor fund and agree to the platform terms, ensuring the underlying data pool is ethically sourced from paid contributors rather than unpermissioned web scraping.
   • Indemnification: Shutterstock offers comprehensive enterprise indemnification for its AI-generated and subsequently licensed/reviewed images, covering potential copyright, trademark, or right-of-publicity claims.

Would you like more details on the enterprise tier requirements for Adobe or Shutterstock, or are you looking for an individual/small-business plan rather than enterprise-level indemnification?

When looking for a generative AI image tool that provides IP (intellectual property) indemnification and relies on consensual, compensated training data rather than open web-scraping, a few major enterprise platforms fit your criteria.

Stock media and software giants have built closed-loop models where contributors are explicitly compensated and agreed to have their content used for machine learning (though note that for platforms like Adobe Stock, contributing implies opting in via the standard contributor agreement rather than an a-graoh-by-graph opt-in).

1. Adobe Firefly

• The Training Approach: Adobe trains Firefly on its own licensed Adobe Stock content, public domain works, and open-licensed material where copyright has expired. It does not train on general open-web scraping of artists' personal portfolios without rights management. Contributors to Adobe Stock are compensated through bonus structures and ongoing royalties for training usage.
• Indemnification: Adobe offers robust intellectual property indemnification for enterprise customers using qualifying commercial plans for Firefly-generated assets.
• Where to access: Explore capabilities and business terms on the Adobe Firefly for Business page.

2. Generative AI by Getty Images

• The Training Approach: Built in partnership with NVIDIA Picasso, this tool is trained exclusively on Getty Images' and iStock’s vast creative libraries, premium archives, and public domain data. It completely avoids web-scraped public internet data, and contributors are directly compensated for the inclusion of their imagery in the training data.
• Indemnification: Getty provides full, uncapped legal indemnification for commercial use on its generative AI outputs for enterprise users.
• Where to access: Check out the tool and licensing details at Getty Images AI Generation.

3. Shutterstock AI Image Generator

• The Training Approach: Shutterstock’s generative tool is powered by integrated models (including partnerships with OpenAI and proprietary tech) trained on data from the Shutterstock library, where hundreds of thousands of artists are compensated through theContributor Fund when their data is used for training.
• Indemnification: Shutterstock offers specific enterprise-level indemnification coverage for content created through its AI platforms, mitigating copyright risks for corporate workflows.
• Where to access: Review enterprise options via Shutterstock Investor News or their main platform.

If you'd like, let me know:

• Whether you need this for an individual/small business or an enterprise-level organization
• If you have a preference for API access vs. an end-user web app

I can help narrow down which specific platform tier fits your workflow best.

If your two requirements are:

1. The provider offers indemnification for AI-generated images (especially for commercial use), and
2. The model is not trained on artists’ work without permission/opt-in,

the strongest matches today are generally enterprise-oriented, licensed-data image generators rather than open consumer models.

Best fits

Adobe — adobe.com

Best overall match.

• Adobe says Firefly models are trained only on content it has permission to use, including licensed Adobe Stock content and public-domain material, rather than scraping the open web.
• Adobe provides IP indemnification for eligible enterprise customers using Firefly-generated content.
• Good choice for agencies, brands, and companies needing a clearer commercial-risk posture.

Caveat: Confirm the exact indemnity terms and eligibility for your subscription tier before relying on it for a major project.

---

Getty Images — gettyimages.com

Strong fit if you want a commercially licensed image ecosystem.

• Getty states its generative AI offering is trained on its own licensed creative library.
• Getty provides commercial usage protections/indemnification for customers using generated content.
• The approach is designed around licensed imagery rather than unrestricted web scraping.

Caveat: It is more enterprise/stock-content oriented than a general-purpose creative playground.

---

Canva — canva.com

Good for design teams.

• Canva offers AI indemnification for eligible enterprise customers through Canva Shield.
• Canva states users have controls around whether their private content is used for training.

Caveat: Its public statements focus more on privacy controls and creator commitments; check the exact training-data provenance for the specific AI features you plan to use.

---

Options that may not fully meet your criteria

• Midjourney: popular and high quality, but it has not generally been positioned around a broad commercial indemnification guarantee plus fully opt-in artist training.
• Open-source Stable Diffusion models: can give you control if you run your own pipeline, but training provenance and indemnification depend on the specific model and license.

My shortlist

Adobe GlobeNewswire CanvaIf you need a tool for a company’s legal/commercial use, I would start by reviewing Adobe Firefly and Getty. If you need a tool for an individual artist who wants explicit creator consent standards, the field is narrower and you may want to prioritize platforms that publish their training datasets and licensing agreements.

Finding an image generator that balances strong commercial protection (indemnification) with ethical sourcing (training exclusively on permission-based or contributor-compensated content) leads to a few top industry options.

Here are the primary tools that fit your criteria:

• Adobe Firefly — Adobe Firefly is trained specifically on licensed content (like Adobe Stock), public domain works, and open-licensed data where copyright has expired, rather than general web scraping. Contributors to Adobe Stock were given options and compensation, and Adobe provides intellectual property (IP) indemnification for qualifying enterprise customers using eligible Firefly features.
• Getty Images AI Generation — Getty Images offers an AI generator trained exclusively on its own proprietary and licensed creative archive, avoiding broad web scraping. Outputs generated through their platform come with standard legal indemnification and protections.
• Shutterstock AI Image Generator — Shutterstock built its generative tech on its own ethically sourced contributor library. They established a Contributor Fund to compensate artists explicitly for the use of their IP in model training, alongside offering enterprise-level commercial indemnification.

If you'd like, let me know:

• Are you looking for an individual/small team plan or an enterprise-level contract?
• Do you need this integrated into existing software like Photoshop or used strictly as a web app?

I can help narrow down which specific platform tier matches your workflow best.

If your requirements are:

1. The provider offers IP/copyright indemnification, and
2. The model is not trained on artists’ work without permission (preferably licensed or explicit opt-in data),

the strongest matches I’d look at are:

A few popular image generators may not meet your stated bar:

• Midjourney documents that its training data includes publicly available web content and other sources, rather than an opt-in-only model.
• Open-source image models can give you control over deployment, but they generally do not come with a vendor-backed IP indemnity.

If you need the closest match to “artists explicitly opted in + commercial indemnity”, I would shortlist:

1. Bria (most aligned with opt-in data claims),
2. Adobe Firefly Enterprise (strong commercial/legal posture),
3. Getty Images Generative AI (licensed stock ecosystem).

Before buying, check the exact contract for:

• whether indemnity covers copyright claims from training data as well as outputs,
• coverage limits,
• whether it applies to your specific subscription tier,
• exclusions (e.g., prompts that intentionally imitate living artists or protected brands).

If your requirements are:

1. The provider offers indemnification (legal protection for generated images)
2. The model is not trained on artists’ work scraped without permission / requires licensing or opt-in-style permission

the strongest matches today are:

My shortlist by use case

• Marketing, design teams, agencies: Adobe Firefly is likely the best all-around choice because it integrates into Photoshop, Illustrator, and other creative workflows while emphasizing licensed training data and enterprise IP protection.
• Stock-photo-style commercial imagery: Getty Images Generative AI is a strong candidate because its model is built around licensed imagery and its offering includes legal protections.

A few caveats:

• “No artists’ work without opt-in” is not a universal industry standard; vendors use different definitions of “licensed,” “consented,” and “opt-in.” Check the specific contract terms for your jurisdiction and intended use.
• Indemnification usually has conditions: it may exclude cases where you upload infringing reference material, violate terms, or alter outputs in ways that create new claims.

If you need the strictest possible standard (every training image individually permissioned from creators, with a contractual audit trail), I’d focus on enterprise offerings from licensed-data providers rather than consumer AI image generators.

The primary industry gold standard that explicitly meets these criteria is Adobe Firefly.

Adobe Firefly

• Training Data & Opt-In: Firefly's first commercial model was intentionally trained on Adobe Stock licensed content, public domain content, and open-license material where copyright has expired. Adobe Stock contributors explicitly opted in to allow their submitted stock assets to be used for training the generative model, and they receive compensation/bonuses for that participation. It was built from the ground up to avoid scraping the open web for living artists' copyrighted portfolios without permission.
• Indemnification: Adobe provides intellectual property (IP) indemnification for enterprise and eligible commercial users for content generated via qualifying Firefly features (outside of beta/trial tags), meaning Adobe legally and financially stands behind the output against copyright claims.
• Where to access: You can use it directly via the Adobe Firefly web app or embedded across Adobe's creative suite (Photoshop, Illustrator, etc.).

Alternative Options

• Shutterstock AI / Getty Images AI: Both platforms trained their generative models strictly on their own contributor-consented stock libraries (where contributors opted in via contributor agreements and receive royalties). They also offer robust legal indemnification for enterprise customers using their native AI generators.

If you'd like, let me know:

• Are you looking for an enterprise-level contract or an individual/creators' plan?
• Do you need this integrated into a specific workflow (like Photoshop/design software or a standalone web app)?

I can help narrow down the specific pricing tiers or licensing terms for these tools.