Bolkra
Sign up now

Effective date: 21 May 2026

Last updated: 21 May 2026

Bolkra Fair Use Policy

1. Why this policy exists

Most acceptable use policies are written to be ignored. They exist so a company can point at them later. This one is different, and it is worth a few minutes of your time, because it tells you what we actually think.

Bolkra was built by people who spent years inside commercial image production. We know what a shoot costs, how long a model release takes to clear, and how a creative director describes the look they want before anyone has lit a single frame. AI image generation changes the economics of that work, and like a lot of change it arrives faster than the rules around it. The questions are genuinely contested. Who owns an image a machine made. Whether the models were trained on other people's work without permission. What happens when a tool can produce a convincing picture of a real person, or a real brand, in seconds.

We could have written a page that says "you must comply with all applicable laws" and left it there. We decided not to. Sophisticated buyers, the agency creative directors and brand-side marketing leads who use Bolkra, deserve to know where a vendor stands before they put a campaign through it. So this policy does two things at once. It sets the rules you agree to follow when you use Bolkra, and it explains our reasoning, including the parts where the honest answer is that the law has not caught up yet.

If you read this and decide Bolkra is not for you, that is a fair outcome. We would rather lose a sale than be vague about something this important.

This policy forms part of our Terms of Service. Breaking it can lead to your account being suspended or terminated, as Section 6 explains.

2. What we do, and do not do, with your data

Start with the simplest and most important commitment, because it is the one customers ask about first.

We do not train AI models on your content. Any of it. We do not use your uploaded reference images, your prompts, or the images you generate to train, fine-tune, improve or develop any AI model. This is true whether you uploaded an image as a product reference, as conditioning for a particular look, or as training data for a character model. It is true of our own systems and it is true of the systems we rely on. Your content is processed for one reason only, which is to produce the output you asked for.

We think this should be the default for any serious B2B tool, and we are aware that it is not. A creative team's reference library and prompt history is competitive information. It should not quietly become training fuel.

Which AI vendors are involved. Bolkra does not run its own foundation models. We use external AI providers to perform image generation and model training. We describe these vendors by category in our Privacy Policy, and the current named list is available on request. We have selected the paid, commercial tiers of these services, on which the vendors commit not to train on submitted content.

The harder, more honest part. The no-train commitment above is about your content. It does not resolve the larger question hanging over this entire industry, which is how the underlying AI models were trained in the first place.

We are not going to pretend that question is settled, because it is not. The large image models that power tools like Bolkra were trained on very large datasets, and the provenance of those datasets is publicly contested. There are active legal cases. There are reasonable people who believe the current approach is fair, and reasonable people who believe it is not. If a vendor tells you this is all resolved and clean, treat that claim with caution.

Here is our position. We chose vendors whose practices are at least transparent enough to evaluate, rather than vendors whose training data is entirely opaque. That is not a perfect answer. It is an honest one. As an industry standard emerges, whether through litigation, regulation, or licensing markets for training data, we intend to move with it, and we would rather tell you that openly than imply a certainty we do not have.

3. What you own

You own the images you generate with Bolkra. As between you and Bolkra, the outputs are yours. You can use them in commercial work, modify them, combine them with other material, and deliver them to your clients. We do not take a commercial cut, we do not license your outputs to anyone else, and we will not use them to promote Bolkra without asking you first.

There is one caveat that we will not gloss over, because it is real and it affects how you should use these images.

Whether an AI-generated image is protected by copyright is an unsettled question, and the answer differs by country. In the United States, the Copyright Office has taken the position that a work generated purely by an AI system, without sufficient human authorship, is not registrable. Other jurisdictions are working through the same question and have not all landed in the same place. The practical implication is this. When you generate an image on Bolkra, you own it in the sense that we claim nothing in it, but the degree of copyright protection that image attracts against the wider world is uncertain and depends on the law where you are and on how much human creative input shaped the result.

We are not your lawyers and this policy is not legal advice. If copyright protection in a specific output matters to a campaign, take advice on it. What we can tell you cleanly is the part we control. Bolkra retains no commercial rights in your outputs and will not use your content to train anything.

We also do not warrant that your outputs are free from third-party intellectual property claims. We cannot, and no honest vendor can. Your outputs are shaped by your prompts and your reference images. You chose those inputs, you know the brief, and you are far better placed than we are to judge whether a given image carries IP risk. That responsibility sits with you, and the Terms of Service set it out formally.

4. Real people in your images

This is one of the areas where the law has teeth, and where we are strict.

If you train a model on a real, identifiable person, or generate the likeness of one, you must have that person's written consent. Not a verbal agreement, not an assumption, not "they are a colleague and they will not mind". Written consent, covering the use you are putting them to. This is a contractual requirement under our Terms of Service. It is not an aspiration and it is not a suggestion.

Bolkra prohibits using the platform to generate the likeness of a real person who has not consented. That includes, and we want to be specific:

  • celebrities and well-known figures, whose fame does not amount to consent;
  • public figures placed in invented or unflattering contexts;
  • private individuals who have not given you permission; and
  • anyone, famous or not, depicted in a way that misrepresents them.

We know how this feature gets used well. An agency trains a model on a paid model who has signed a release, or on a member of staff who has agreed in writing, so that a campaign can be extended without booking a full reshoot. That is a legitimate, valuable use, and it is exactly what the consent requirement protects. The release you would obtain for a traditional shoot is the same discipline. Apply it here.

For agencies. If you use Bolkra to train models on your own staff or on paid models, you are responsible for obtaining the consent, for keeping a record of it, and for staying within its scope. If consent is withdrawn, you must stop using the model and delete it. You indemnify Bolkra against claims that arise from inadequate or absent consent. Your contracts with the people you photograph, and with your own clients, should reflect this.

On enforcement, we will be honest. We cannot inspect every output, and Section 6 explains why. We do not pre-screen your generations for unauthorised likenesses. What we do is reserve the right to investigate when something is reported or flagged, and to suspend or terminate accounts that produce unauthorised likenesses of real people. If you are using Bolkra to make deepfakes, this is not the tool for you, and we will act on it when we find it.

5. What we will not generate

The sections above explain our reasoning at length because the questions are genuinely hard. This section is short, because these lines are not hard at all.

You may not use Bolkra to create, upload, store or share:

Child sexual abuse material. This is an absolute prohibition. It is not a fair use question, it is not a grey area, and there is no context in which it is permitted. We use automated detection for this category, we act immediately, we terminate without refund, and we report to the relevant authorities and child safety organisations.

Imagery that facilitates violence, terrorism or hate. No content that promotes, incites or assists violence, terrorist activity, or hatred and dehumanisation directed at people because of who they are.

Imagery that infringes third-party intellectual property. No content that infringes someone else's copyright, trademark, design rights or other IP where you do not hold the necessary licence or permission, including passing off or misleading use of third-party brands.

Deepfakes and deceptive imagery. This connects to Section 4 and is worth stating in its own right. You may not use Bolkra to create:

  • imagery designed to deceive viewers about the identity of a real person depicted;
  • imagery of real public figures placed in fabricated political, sexual or criminal contexts;
  • imagery designed to be passed off as photographic evidence of events that did not happen; or
  • imagery that infringes a person's right of publicity or invades their privacy.

Imagery depicting or facilitating illegal activity. No content that depicts, promotes or helps carry out activity that is unlawful.

This is not an exhaustive list of every bad idea. Use judgement. The spirit of the policy is that Bolkra is a professional production tool, and it should be used the way a professional would use a studio, a camera and a model release: with the consents in place and without trying to deceive anyone.

6. How we enforce this

We will be straight with you about how enforcement actually works, because a policy that pretends to do more than it can is not worth much.

We cannot pre-screen every prompt and every output. No platform operating at the scale of real commercial production can, and any vendor that claims otherwise is overstating it. Bolkra is built for sophisticated B2B customers running professional workflows, not for anonymous consumer-facing image generation, and our enforcement model reflects that.

In practice, enforcement has three layers.

Automated detection for the hard prohibitions. We use automated systems to flag content in the categories where the harm is most serious and least ambiguous, child safety first among them. These systems are not perfect, but for that category they are essential and they run regardless of plan or tier.

Customer compliance as the primary mechanism. For everything else, the consents, the IP judgements, the brand boundaries, the primary safeguard is you. You are a professional, you know your brief, and you are in the best position to keep your work inside this policy. That is why this policy explains its reasoning rather than just listing rules. We want you to agree with it, not just acknowledge it.

Investigation and reporting. Where misuse is reported to us, or surfaces through our systems, we investigate. You can report a concern using the contact details in Section 8.

Consequences. If you breach this policy, we may suspend or terminate your account under our Terms of Service. For serious breaches, including anything involving child safety, unlawful content, or non-consensual likenesses, we may act immediately, without notice, and without a refund. We would always rather a customer stayed inside the policy than face enforcement. But where a breach is serious, we will not hesitate, and the commercial value of an account will not change that decision.

7. Changes to this policy

This is version 1.0 of the Fair Use Policy. As the law develops, as the AI vendor landscape shifts, and as we learn from how the platform is used, this policy will change. That is expected. It would be a worse sign if it never did.

When we update this policy, we will change the version number and the "Last updated" date at the top. For a material change, we will give reasonable notice by email or through the platform before it takes effect. Older versions can be made available on request, so you can see how our position has moved over time. We think that history is part of being accountable for the stance we take.

8. Contact us

If you have a question about this policy, want to report misuse, or want to talk through a use case before you commit a campaign to it, we would genuinely rather hear from you first.

This Fair Use Policy forms part of, and should be read together with, the Bolkra Terms of Service and the Bolkra Privacy Policy.