Carriers Are Testing AI Clones That Answer Calls in Your Voice
A new wave of AI voice clone technology is moving from apps into the phone network itself. REALLY, a mobile virtual network operator (MVNO) running on T-Mobile’s infrastructure, is building “Clone,” an AI assistant designed to sound like you and answer calls on your behalf. After training on your voice and communication style, Clone can pick up incoming calls, talk to the caller in a voice that mimics yours, and then send you a summary of what was discussed. The company pitches it as a way to handle dreaded tasks such as rescheduling appointments, dealing with customer support, confirming bookings, and filtering unwanted or spam calls. Unlike standalone AI apps, however, this system sits directly inside your phone service, meaning your carrier becomes the custodian not just of your call records, but of a digital replica of your voice and conversational habits.
What AI Voice Clones Can Do—and Why People Want Them
The appeal of carrier-level AI phone calls is obvious: most people would gladly offload tedious or awkward conversations. REALLY’s Clone aims to answer unknown numbers, figure out why the caller is reaching out, and handle simple tasks end-to-end. It can negotiate with customer service, manage bookings, and even intentionally keep scammers on the line to waste their time, all while presenting itself in a voice tailored to sound like you. Supporters argue this is a natural evolution from AI that already summarizes emails, manages calendars, and drafts messages. Embedded in the network, an AI assistant could act as a real-time call screener and personal secretary. For users with call anxiety or heavy administrative burdens, phone carrier voice cloning promises a frictionless buffer between them and the endless stream of low-priority calls—without having to install or manage a separate app.
Voice Clone Privacy Risks: Who Owns Your Digital Voice?
Embedding AI voice clones in carrier services creates serious voice clone privacy risks. To build a convincing replica, the system must capture and store samples of your speech, plus metadata about how you communicate—tone, pacing, and preferred phrases. Unlike an independent app, a carrier-level clone means your phone company gets direct access to this voice model, alongside your phone number and call history. That concentration of data makes any misuse or breach far more consequential. Critics note that carriers have already faced scrutiny over privacy practices, including allegations involving screen recording. AI voice clone technology also tends to funnel user data into company-controlled storage, where it can be processed, retained, and potentially shared with advertisers or third parties. Once a lifelike voice model exists, it’s difficult for consumers to know how long it will be kept, whether it can be fully deleted, or who exactly can trigger it to speak on their behalf.
Voice Impersonation Security: When Your Clone Talks Like You
Beyond privacy, AI phone calls powered by voice clones introduce new security and impersonation threats. A high-quality model that convincingly mimics your speech could be misused if attackers gain access—whether by breaching a carrier system, exploiting weak authentication, or manipulating AI safeguards. With many organizations still relying on voice-based verification for customer support, an attacker wielding your clone could potentially pass security checks, alter accounts, or authorize transactions. Even if carriers build on decentralized networks or tout advanced security, researchers have repeatedly shown ways to manipulate AI systems and extract sensitive behavior. The stakes are higher when the asset at risk is your own voice, tied to your phone number and identity. Without clear, robust controls—such as strict opt-in, hardened authentication, and limits on what the clone is allowed to do—the line between helpful automation and dangerous voice impersonation security gaps becomes dangerously thin.
What Consumers Should Demand Before Opting In
Telecom regulation has not yet caught up with carrier-operated voice clones, leaving many questions unanswered: who is legally responsible if your AI double is abused, how long data is retained, and what counts as valid consent for using your voice model. Until clearer rules emerge, consumers need to scrutinize any offer to let a carrier “act on your behalf.” At minimum, you should expect explicit disclosures about how your voice is collected, where it is stored, whether it is encrypted, and how you can revoke access or delete your model. You should also look for fine-grained controls—such as restricting the clone to spam handling only, or blocking it from accessing sensitive accounts. While the promise of fewer spam calls and automated admin chores is tempting, the power imbalance is real: once your phone carrier voice cloning system has your digital voice, reclaiming control could be far harder than saying “no” at the start.