Acceptable Use Policy

Consent means the person knew their voice or likeness would be synthesised, understood roughly what it would be used for, and agreed. Consent given for one production does not extend to another. Consent from a rights-holder is not the same as consent from the performer where the law requires the performer’s own — and in a growing number of jurisdictions it does.

You are responsible for holding that evidence. We may ask you for it if a complaint is made about content you processed.

• Non-consensual voice or likeness. Voice clones, dubs, or lip-synced video of a real person made without their consent, including public figures.
• Deceptive synthetic media. Output presented as an authentic recording of something a person never said or did, or engineered to make viewers believe it is genuine.
• Election and civic manipulation. Synthetic depictions of candidates, officials, or electoral processes intended to mislead voters, or any use prohibited by election-integrity or online-harms legislation in the jurisdictions you target.
• Fraud and impersonation. Content intended to impersonate a person or organisation to obtain money, credentials, or access — including voice used to defeat voice-based authentication.
• Sexual content involving real people without consent, and any sexual content involving minors. This includes non-consensual intimate imagery and any synthetic equivalent. We report child sexual abuse material to the appropriate authorities.
• Harassment, hate, and incitement. Content that targets a person or group for abuse, or that incites violence, self-harm, or unlawful discrimination.
• Infringing content. Media you do not hold the rights to localize, including music, performances, and footage licensed on terms that do not permit derivative or dubbed versions.
• Unlawful content. Anything otherwise illegal where you operate or where the output will be published, including defamatory material and content that breaches court orders or sanctions.
• Restricted personal data. Media containing government identification numbers, payment card or financial account data, health records, biometric identifiers used for identification, or personal data of children under 16, unless we have agreed in writing in advance.

• Reverse-engineering, decompiling, or attempting to derive the source code, models, prompts, or pipeline design behind the service.
• Reselling, sublicensing, or providing the service to third parties as your own, or using it to build a competing product.
• Circumventing metering, quotas, rate limits, or billing, including by registering accounts to evade limits or to abuse trial credit.
• Scraping, crawling, or bulk-extracting from the service beyond what the interface offers, or introducing malware or malicious code.
• Probing, scanning, or load-testing our infrastructure, or attempting to access another customer’s account or content, without our prior written permission.
• Interfering with the service’s operation or with other customers’ use of it, including through disproportionate load.

Several jurisdictions now require AI-generated or AI-manipulated audio and video to be labelled, watermarked, or disclosed — with stricter rules in election, news, and advertising contexts. The EU AI Act’s transparency obligations are one example; state-level digital-replica and voice laws in the United States are another; Singapore’s online-harms and election-advertising legislation is a third.

Where any such rule applies to your output, you must comply with it. We recommend disclosing that a dub is AI-generated whether or not the law requires it. Dubsmith does not determine what applies to your publication and gives no assurance that any output is lawful to publish.

We may investigate suspected breaches and take action proportionate to what we find: removing content, restricting a feature, suspending processing, or terminating the account. Where a breach is serious, ongoing, unlawful, or poses a risk to someone, we may act immediately and without prior notice, and may preserve and disclose material to law enforcement where we are required to or reasonably believe it necessary.

No refunds for enforcement action. Suspension or termination for breach of this policy does not entitle you to a refund of fees paid or a waiver of amounts owed.

No duty to monitor. We do not pre-screen content, and nothing here obliges us to. Content passing through the service is not an indication that we have reviewed it or consider it permitted.

If you believe content processed through Dubsmith uses your voice or likeness without consent, infringes your rights, or otherwise breaches this policy, email contact@dubsmith.io with enough detail to identify the content and your claim. We take these reports seriously and act on them.

We may update this policy as the law and the risks change; the current version is always the one on this page, and material changes carry a new date.