Policy Statement on the Use of AI Voice Cloning Technology

What is Voice Cloning?

Voice cloning is creating a digital replica of a person's voice using AI to mimic their unique speech patterns, tone, and cadence, allowing a computer to generate new speech in that identical voice from text input, useful for creative projects and assistive tech but also ripe for scams like impersonating family or executives for fraud. 

Prohibition on Voice Cloning

SSAFA prohibits the creation, upload, or use of AI‑generated voice clones of any staff member, volunteer, beneficiary, or external party.
Voice cloning technology must not be used for internal or external communications, marketing, training, media production, or experimentation.

GDPR & Legal Basis

Voice data is considered biometric personal data under Article 4(14) and Article 9 of the GDPR. Processing biometric data for identification or replication requires:

• Explicit, informed, freely given consent
• A clear, specific purpose
• The ability to withdraw consent at any time
• Proof that all data and models can be deleted when consent is withdrawn

Voice‑cloning tools cannot reliably meet these requirements. As such, SSAFA cannot lawfully process voice data using these technologies at this time.

Consent & Power Imbalance

GDPR guidelines state that consent is not valid if there is a power imbalance (e.g., employer to employee). Employees may feel pressured to agree, which invalidates the consent and exposes SSAFA to compliance risk. Therefore, consent‑based processing is not appropriate or acceptable for voice cloning.

Ethical & Reputational Risks

Voice cloning presents serious ethical risks:

• Potential creation of misleading or harmful content
• Loss of trust from staff, volunteers, beneficiaries, and supporters
• Very high risk of impersonation or misuse
• Public perception of manipulation or dishonesty

As a charity that relies on credibility and trust, SSAFA must avoid technologies that could undermine confidence in its communications.

Alternatives

SSAFA encourages the use of:

• Standard text‑to‑speech voices from approved M365 tools
• Human‑recorded audio
• AI‑generated content that does not replicate real individuals, such as the service provided https://artlist.io that we are subscribed to.

These alternatives avoid biometric processing and remain compliant with GDPR.

Review

This policy will be reviewed annually or sooner if regulatory or technological developments require.