Security

We protect individual accounts and creative content with encryption, role-based access, monitoring and operational safeguards. User content is private by default. Text generation may use Azure OpenAI models sold by Azure or Anthropic Claude through Microsoft Foundry as model versions change. Microsoft says Azure-sold model prompts and outputs are not used for foundation-model training without customer permission; Anthropic's commercial terms separately prohibit training on Claude customer content. Safety review may still apply under the selected provider's terms. Kie.ai has confirmed by email that it does not use prompts, uploaded inputs or outputs for training in its own handling, while upstream providers follow their own terms; ElevenLabs' terms make training for audio, voice and music/sound content opt-out-based, and Kraftflix has disabled its account's "Improve the models for everyone" setting.

Data protection

• Encryption in transit and at rest for project data, media assets and backups.
• Private-by-default projects — generated outputs are not public unless a user chooses to share them.
• Kraftflix does not train its own AI models on your content. For text generation, Microsoft says prompts and outputs sent to Azure OpenAI models sold by Azure are not used to train foundation models without customer permission. Anthropic operates Claude in Microsoft Foundry and its separate commercial terms prohibit training on customer content, including when Claude is hosted on Azure. Kie.ai has confirmed its own no-training practice by email, but its published terms do not include that assurance and upstream-provider terms remain separate. ElevenLabs training for audio, voice and music/sound is opt-out-based; Kraftflix has disabled its ElevenLabs account's Data use setting. See the Privacy Policy.

Access & identity

• Role-based access control and restricted administrator access.
• Least-privilege internal access to customer data, with access logging.

Infrastructure & monitoring

• Separate user-data environments, backups, vulnerability patching, incident logging, and access monitoring.
• Script, text and prompt-planning content may go to Azure OpenAI models sold by Azure or to Anthropic Claude through Microsoft Foundry; the selected model's terms, safeguards and deployment apply. Scene-level image and video prompts go to Kie.ai, which says it does not use prompts, uploaded inputs or generated outputs for training in its own handling; paid upstream-provider accounts follow their applicable terms. Audio, voice and music/sound content goes directly to ElevenLabs, where training is opt-out-based and Kraftflix has disabled its account's Data use setting. Razorpay handles payments; some AI processing occurs outside India.

Incident response

We maintain an incident-response process and will assess, contain, investigate, log and remediate security incidents. Where required by applicable law, we will notify affected users and relevant authorities.

Support & availability

For support or security questions, contact us. We acknowledge privacy grievances within 2 business days and target resolution within 30 days, as described in our Privacy Policy. Where required, we notify affected users about material incidents.

Current AI provider practices

For Azure OpenAI models sold by Azure, Microsoft says prompts and outputs are not used to train, retrain or improve foundation models without permission; content filtering and limited review of flagged material may occur. For Claude through Microsoft Foundry, Microsoft explains that Anthropic is the seller, operator and independent data processor even when Claude is hosted on Azure. Anthropic's commercial terms say it may not train on customer content; its safety systems and limited review of flagged content may apply. Claude may be hosted on Azure or Anthropic infrastructure depending on deployment, and model versions may change over time. Kie.ai receives scene-level image and video prompts, not full scripts. It confirmed to us by email that it does not use prompts, uploaded inputs or generated outputs for training or fine-tuning in its own handling. Its published terms do not include that assurance, and paid upstream providers follow their own applicable terms. ElevenLabs receives audio, voice and music/sound content directly and allows training unless the account opts out; Kraftflix has disabled that setting for its account. The opt-out is not retroactive. See the Privacy Policy for data-sharing and transfer details.

Kie.ai reports 14-day generated-file retention and two-month log and task-record retention, including prompt text in those records, as current operational practices rather than contractual guarantees. It has not offered a separate DPA, enterprise agreement, SLA, dedicated support or third-party intellectual-property indemnity. See our Privacy Policy for deletion limits and our Terms of Service for output rights.

Report a vulnerability

Found a security issue? Email support@kraftflix.com with details and, if possible, steps to reproduce. We investigate every report, ask that you give us a reasonable window to fix an issue before public disclosure, and won't pursue legal action against good-faith, non-destructive testing reported to us privately.