StudioNeural: AI Ethics Statement

Financial Year 2026 | Published: February 2026

Risk Of Misuse

One of the primary concerns surrounding our AI models and products - particularly our NerualMotion, NeuralSync, NeuralCine and NeuralVoice systems, is the risk of these technologies being misused to create fraudulent, deceptive, or harmful content. For example, face-veiling technology could be used to produce videos that impersonate individuals without their consent, while voice cloning could replicate someone's voice for fraudulent purposes, such as impersonating them in financial or legal transactions.

To address this, we have implemented strict access controls and work only with carefully vetted clients. Each potential client undergoes a thorough review process to ensure that their intended use of the AI is both legitimate and ethical. We assess their use case, objectives, and the context in which the technology will be applied. Only clients with clearly documented, ethical use cases are granted access to our services/models. This reduces the risk of misuse and ensures that the technology is deployed only in environments where it serves a constructive purpose, such as in the entertainment or media industries for post-production work.

Additionally, every AI output is subjected to internal review by a dedicated team of experts. These outputs are examined to ensure compliance with our ethical guidelines, and any potential risks or anomalies are flagged. If any content is found to be inappropriate, non-consensual, or misaligned with ethical standards, it is immediately rejected, and the client is contacted for further investigation. This review process adds a critical layer of protection, ensuring that our AI outputs are always created within the bounds of legal and ethical responsibility.

To further mitigate misuse, we regularly monitor and track how the AI models are being used. This approach ensures that we maintain strict control over the use of our technology at all times.

Privacy Concerns

Given that our AI models can manipulate sensitive personal information, such as an individual's face or voice, there is an inherent risk of violating privacy rights.

To address these privacy concerns, we operate on a strict consent-based framework and zero retention "opt-out" from model training for all client data containing biometrics and likeness (audio and image). This means that no content is ever produced without the explicit consent of the individuals involved. For instance, in cases where a face-swapping or voice cloning is required, written and documented consent is obtained from the person whose likeness or voice is being used and/or from rights holder(s). If we are creating AI-generated content for actors, voice-over artists, stunt doubles or public figures, we ensure that the necessary agreements are in place to comply with their intellectual property and privacy rights.

We also recognise that there may be some exceptions, such as when we are working on parody content. In such cases, we consult with legal experts and our clients to navigate the nuances of fair use and ensure that the content complies with applicable laws. These discussions are held with full transparency, and any decisions made are thoroughly documented to protect both our company and the individuals involved.

Our commitment to privacy also extends to the way we store and process data. All data, whether image or audio, is handled in a secure, encrypted environment that adheres to industry best practices for data protection and privacy compliance, including ISO27001 and the General Data Protection Regulation (GDPR). By employing robust security measures, we ensure that personal data is safeguarded at every step of the pipeline and process.

Bias and Discrimination

A well-documented risk in AI systems is that models, if not properly trained, can produce biased outputs that unfairly impact certain groups based on gender, race, ethnicity, or other demographic factors. This issue is particularly prevalent in AI models that rely on large-scale datasets, as training data may contain inherent biases that skew the model's behaviour.

To counter this, we have implemented a comprehensive bias mitigation strategy. First, we ensure that our training data is ethically sourced from peer-reviewed academic sources and representative of a wide range of demographics. For the face-veiling models, we have sourced images that reflect a diverse set of faces in terms of age, gender, ethnicity, and facial structure, ensuring that no group is underrepresented or overrepresented. Similarly, for our voice cloning model, we have incorporated peer-reviewed audio data from speakers with varying accents, languages, and speech patterns to prevent the model from favouring a particular demographic.

We conduct regular bias reviews during and after the training process for our models. These reviews involve assessing the model's outputs across different demographic groups to identify any signs of bias or discrimination. If biases are detected, we retrain the model with adjusted datasets to ensure more equitable performance. This proactive approach helps minimise the risk of biased outputs and ensures that our AI models deliver fair and accurate results for all users, regardless of their background.

Legal and Ethical Compliance

Compliance with local and international laws is central to the responsible deployment of our AI models. Our team works closely with legal experts specialising in privacy, media, and intellectual property law to ensure that all uses of the AI models align with relevant regulations.

We also require clients to sign legal agreements that outline the permissible uses of the AI products. These agreements make clear that any misuse of the technology, such as creating non-consensual or fraudulent content, will result in immediate termination of access to our models and potential legal consequences.

Societal Impact

We understand that the potential impact of AI-driven technologies like face-swapping and voice cloning extends beyond individual use cases and can have broader societal implications. For example, the misuse of these technologies could lead to a decline in public trust in media, as face swapping and voice clones may be used to manipulate or deceive audiences. This could fuel disinformation, especially in political or social contexts, and lead to significant harm for those targeted by such content.

To mitigate these societal risks, we actively engage in public awareness campaigns and client education about the ethical use of AI. We work with our clients to ensure they understand the potential consequences of misuse and encourage them to adopt responsible AI practices. This includes promoting transparency in how AI-generated content is labeled and ensuring that audiences are aware when they are viewing or listening to content created using AI technology.

We also collaborate with regulatory bodies and industry organisations to help shape guidelines and best practices for AI-driven media manipulation technologies. By working within a broader ecosystem of ethical AI practitioners, we contribute to the development of standards that ensure the responsible and accountable use of these technologies in society. Additionally, we advocate for legislative and policy measures that address the challenges posed by AI misuse, helping to create a safer and more secure environment for AI innovation.