What you may not make.

This policy covers every prompt, reference photo, persona, image, video and caption created, uploaded, stored or distributed through Personifex. It applies to workspace owners and to every member they invite: an owner is responsible for their members’ use.

It forms part of the Terms of Service. Breaching it is a breach of those terms, and is grounds for removing content or ending your access.

Personifex makes convincing images of people who do not exist. That is a useful tool and also a dangerous one, and the whole point of this page is to be specific about where the line falls rather than hiding behind “don’t do anything bad”.

Two categories have no exceptions, no context that makes them acceptable, and no warning step before enforcement.

Child sexual abuse material. You may not create, upload, request, store or distribute any sexual content involving a minor, or any content that sexualises a person who is or appears to be under 18, whether real, synthetic, stylised, or claimed to be a fictional adult who looks younger. This includes reference photos of minors used in any sexual context.

We terminate immediately and without notice, we preserve what the law requires us to preserve, and we report to the relevant authorities. There is no appeal on this ground.

Non-consensual intimate imagery. You may not create or distribute sexual or intimate imagery of a real, identifiable person without their explicit, informed, documented consent. This covers “undressing” a clothed photo, placing someone’s face on sexual content, and generating intimate imagery of an ex-partner, a colleague, a public figure or a stranger. Consent to be photographed is not consent to this.

Reports of non-consensual intimate imagery from the person depicted go to the front of the queue. See section 10.

A persona must be fully synthetic, or built on someone who agreed to it. There is no third option.

• Every reference photo you upload must depict a person who gave informed consent to their likeness being used to generate synthetic images, or must not depict a real person at all.
• You must hold evidence of that consent and produce it if we ask. We may ask at any time, and we may act on a report while waiting for an answer.
• Consent must be specific. Someone agreeing to appear in a photograph has not agreed to be turned into an AI persona posting indefinitely.
• You may not build a persona on a public figure (celebrity, politician, influencer, executive) without their agreement. Fame is not consent.
• You may not use the likeness, name or voice of a deceased person where their estate or local law restricts it.
• A person depicted in a persona may withdraw consent. If they contact us and you cannot show valid consent, the persona and its images come down.

This is the obligation we enforce most readily, because it is the one where the person harmed has no relationship with us and no other route to being heard.

A synthetic person is not inherently deceptive. Using one to make someone believe a false thing to their detriment is. You may not:

• Present synthetic content as an authentic photograph or recording of a real event or a real person.
• Impersonate a real person, a business, or an official body, including creating an account or persona that a reasonable person would take for someone else.
• Run romance, investment, crypto, or advance-fee scams, or any operation that uses a fabricated person to build trust before extracting money. Catfishing an individual is covered here whether or not money changes hands.
• Create fake reviews, fake testimonials, fake endorsements, or fabricated before-and-after results.
• Produce political or election content that misrepresents a candidate, a party or a voting process, or that is designed to suppress or mislead voters.
• Present a synthetic persona as a qualified professional giving medical, legal, financial or psychological advice, or as a regulated practitioner of any kind.
• Manufacture engagement: bot networks, coordinated inauthentic behaviour, or personas whose purpose is to make an opinion look more widely held than it is.

• No content that harasses, bullies, threatens, stalks, defames or degrades a person or group, and nothing that constitutes a campaign against an individual.
• No content that promotes or glorifies violence, terrorism, or a violent extremist organisation, or that uses hate symbols or slurs against a protected characteristic.
• No content that encourages self-harm, suicide, or disordered eating, or that presents them as aspirational.
• No doxxing: publishing someone’s address, workplace, phone number, or other private identifying information.
• No content designed to exploit or endanger children in any way, including grooming behaviour and content that sexualises childhood generally.
• Nothing illegal where you are or where we are, and nothing that infringes copyright, trademark, or other rights of a third party.

• Do not attempt to break, probe or circumvent the service’s security, access another workspace’s data, or bypass plan limits.
• Do not scrape the service, resell access, or run it as a white-label product for others without our agreement.
• Do not use someone else’s API key, or a key you are not authorised to use.
• Do not upload malware, or use generated content to distribute malware or phishing.
• Do not use a model outside its own licence. Your provider hosts models from several vendors, and the licence attaches to the model you pick, not to the provider you signed up with; two models on the same account can permit different things.
• Do not use the service to build a dataset for training another model where that breaches your provider’s terms.
• Do not evade an enforcement decision by creating a new workspace or account.

What you make here, you publish elsewhere, and the rules that apply there are yours to follow, not ours.

Every platform has its own policy on AI-generated content, synthetic personas and disclosure, and they change often. Instagram, TikTok and X each have labelling requirements, and some prohibit undisclosed synthetic personas outright. Advertising and consumer-protection law may separately require you to disclose that a person shown is not real, particularly in a paid promotion or an endorsement.

Complying with all of that is your responsibility. We do not check it, we cannot check it, and a Personifex subscription is not evidence that anything you publish is permitted where you publish it. If you are operating in a jurisdiction with AI transparency rules, take advice on them.

Presenting synthetic content as a genuine depiction of a real person is prohibited by section 4 regardless of what any platform allows.

You should know exactly how much screening stands between you and a mistake, so: we do not pre-screen anything.

Personifex runs no automated content classifier, no image scanning, no hash matching, no prompt blocklist and no moderation queue. Your prompts and your generated images are not inspected by us before or after generation.

The AI providers you connect apply their own filters, which are theirs and which we do not audit. One exception you should know about: for the optional caption and hashtag features, when the provider is Google Gemini, we set the adjustable safety thresholds to their most permissive setting so it writes ordinary flirty or aesthetic marketing copy instead of refusing. Google still applies limits that this cannot override, and it does not affect image or video generation at all. But it means we have loosened a filter rather than tightened one, so do not treat caption output as safety-screened.

We act when something is reported to us, or when we become aware of it another way. That means the responsibility for what you generate sits with you, and it means the absence of an error message is not approval. Nothing generating successfully implies that it complies with this policy.

Email abuse@personifex.com. Anyone may report; you do not need an account, and you do not need to be the person affected.

If you are reporting intimate imagery of yourself posted without your consent, say so in the first line. Those go to the front of the queue and we will not make you prove ownership of the likeness before we act to limit further harm.

To act quickly we need:

• Where it is: a URL, or enough detail to identify the specific content or account.
• Why it breaches this policy or the law, in your own words. A sentence is enough.
• How to reach you: a name and email address, so we can acknowledge and tell you the outcome. You may report child sexual abuse material anonymously.
• A statement that you believe the report is accurate and made in good faith.

We will confirm receipt, and we will tell you what we decided and why. Reports made in bad faith or at volume to harass someone will be disregarded.

If you are a rights holder reporting infringement, include what the work is and confirm you are authorised to act for the owner.

We are a small team, so these are deliberately the commitments we can actually keep rather than the ones that sound best.

• Acknowledgement: within 24 hours. Every report gets a reply confirming we have it.
• Child safety and non-consensual intimate imagery: within 48 hours. Confirmed cases are removed within that window, and usually far sooner. Child sexual abuse material is actioned as fast as we physically can, without waiting on any process.
• Impersonation, likeness without consent, and fraud: within 72 hours.
• Everything else: within 7 days, including intellectual property disputes, which usually need input from both sides.

If something needs longer (a genuinely contested factual question, or a legal opinion), we will tell you that within the window rather than let it go quiet.

We match the response to the breach. Where a breach is capable of being fixed and nobody is being harmed in the meantime, we would rather tell you than punish you.

• A warning, explaining what breached the policy and what to change.
• Removing the specific content: an image, a persona, a reference photo.
• Restricting the workspace. You keep access to everything already in it (signing in, browsing, downloading and exporting all still work), but it cannot be used to make anything new: no generating, no uploading, no writing prompts and no scheduling posts.
• Terminating access for serious, repeated, or deliberate breaches.

The two categories in section 2 skip the ladder. Child sexual abuse material means immediate termination and a report to the authorities. Confirmed non-consensual intimate imagery means immediate removal and a review of the whole workspace.

We may also act where the law requires it, or where a court or competent authority orders it. Evading an enforcement decision with a new account is itself a ground for termination.

If we remove your content or restrict your access, we will tell you, by email to your account address, or by a notice inside the workspace itself, or both, and the notice will set out:

• What we did, and to which content or account.
• The facts we relied on, and whether the matter came to us as a report or another way.
• Which part of this policy or which law it breached.
• Whether any automated tool was involved in the decision. Today the answer is always no: every enforcement decision here is made by a person.
• How to appeal, and by when.

The only exception is where telling you would be unlawful, or would obstruct an investigation into child sexual abuse material.

If you think we got it wrong, say so. Reply to the enforcement notice or email abuse@personifex.com within 14 days, and tell us why the decision was mistaken.

Appeals are free, and are reviewed by a person, not by the same automated process, because there is no automated process. We aim to decide within 14 days and will tell you the outcome with reasons. If we agree with you we restore the content or the access and say so plainly.

If you reported something and we declined to act, you can appeal that too, on the same terms. The route is not only for the person who was actioned.

There is no appeal against enforcement for child sexual abuse material.

Our decision on an appeal does not remove your right to go to a court, or to a dispute resolution body, or to complain to a regulator where one applies to you.

We respond to valid legal process from a competent authority. We check that a request is properly issued and appropriately scoped, and we push back on ones that are not.

Where we are permitted to tell you about a request concerning your account, we will. Where we are legally barred from doing so, we will not.

We report child sexual abuse material to the relevant authorities and preserve what the law requires. Otherwise, content removed under this policy follows the normal deletion path described in the Privacy Policy. We do not keep a private archive of removed material beyond what a legal obligation requires.

Legal process should go to legal@personifex.com.

We will update this policy as the product and the law change. Material changes are announced under the process in the Terms of Service. Where a change closes a genuine gap (a new category of abuse we had not anticipated), it may take effect immediately.

Report abuse: abuse@personifex.com. Questions about this policy, or legal process: legal@personifex.com. Privacy requests: privacy@personifex.com.

If you are in immediate danger, contact your local emergency services. We are a software company and we cannot help with an emergency.

Version 2026-08-14