- Photographer: Zainah Alagbari
The voice on the podcast is familiar. The intonations, the pauses, the slight catch of breath between sentences, all recognizable. But, the speaker has never been in a recording studio. Their voice is being cloned without consent.
A voice is more than just sound; it is as unique as a fingerprint. Yet, with the rapid rise of artificial intelligence in audio generation, voices are increasingly vulnerable to replication and misuse.
Kristian Hammond, professor of computer science at Northwestern University and co-founder of AI company Narrative Science, told The Caravan that voice cloning is “a machine learning approach where a system, at its core, generates voice and then undergoes secondary training with an individual’s voice.”
McAfee, an American company focused on online protection, published a report titled Beware the Artificial Impostor in May 2023. According to its lab’s security researchers, only three seconds of audio are enough to clone a person’s voice.
The company surveyed over 7,045 people between April 13 and April 19, 2023, across seven countries, including the United States, United Kingdom, Germany, France, India, Japan, and Australia, and found that a quarter of adults had experienced some form of AI voice fraud.
The survey showed that with 53% of adults unknowingly sharing their voice data online every week through social media, voice notes, or other digital platforms, a new weapon has emerged in the cybercriminals’ arsenal: AI voice cloning.
This increasing impact of voice cloning technology brings into focus the role of legislation in regulating its misuse.
Mervat Abou Oaf, professor of practice in the department of journalism and mass communication and an expert in media law and policy, highlights the gap in regulations.
“We do not have laws that evolve with technology,” she said.
Despite this delay, some legislation and laws exist to protect human vocal identities. People need to be aware of them to prevent their voices from being cloned, replicated, or stolen.
The most prominent laws regulating this technology, according to Abou Oaf, are the Telecommunications Regulation Law and the Cyber Crime Law.
The former has been in effect since 2003. It prohibits recording or transmitting someone’s voice without permission. Then in 2015, the Cyber Crime Law came into effect putting provisions on identity theft and impersonation.
Falsely claiming ownership of a sound or voice can also be considered piracy. Abou Oaf emphasized Law No. 82 of 2002 on the Protection of Intellectual Property (IP) Rights, which lays down both maximum and minimum penalties for such violations.
To further protect individuals’ voices from being used by AI, Mahmoud Othman, entertainment lawyer and founder of Othman Media Law Firm, referred to Egypt’s National Intellectual Property Strategy.
This strategy necessitates that all intellectual property laws in the country will be amended within five years to accommodate the advancements in technology.
“Three years have passed of these five years,” Othman told The Caravan, “two years remain, and the state still has a huge responsibility to clarify the rights related to voice.”
According to Abou Oaf, the use of human voices without consent and its legal consequences depend on the purpose of its use.
This could include harm, personal interest, entertainment, or other reasons.
“Egyptian law is more concerned with the image than the sound, because it is considered a copyright entitlement, but the sound is a related right,” Othman said.
He also stressed that the biggest challenge facing lawmakers in regulating AI-generated audio is its mystery. The limits of artificial intelligence are yet to be fully known, so the question of whether AI or humans owns the rights remains prevalent.
But this hasn’t prevented some from taking advantage of the confusion.
“Telecommunications companies have attempted to use artist voices with the help of artificial intelligence to generate Interactive Voice Response (IVR) for customers,” Othman explained.
To address this, voice performers must be afforded legal protection beginning with a contract. Othman emphasized that agreements must explicitly prohibit voice cloning or replication.
He also confirmed that, in his capacity as a lawyer, he follows this procedure with many voice actors and artists working in advertisements.
Meanwhile, others see potential in AI’s ability to disseminate information, if used responsibly.
Kim Fox, professor of practice in the department of journalism and mass communication and an expert in audio production, considers this a promising development. She notes that if the information is accurate, the identity of the source becomes less significant.
“But transparency, disclosure, and ensuring people have access to the information should be the priority,” said Fox.
As for the steps that governments should take to prevent AI technologies from exploiting human voices, Abou Oaf suggested,“Review, renew, and match the speed with which technological evolution is operating.”
According to Counselor Mahmoud Fawzy, Minister of Parliamentary and Legal Affairs and Political Communication in Egypt, the government is drafting a law to regulate artificial intelligence use across various sectors and set ethical guidelines.
The process entails the amendment of some provisions of Law No. 175 of 2018 on Combating Information Technology Crimes to impose stricter penalties on the establishment, management, promotion, or practice of electronic sites. The ethical considerations of OpenAI use in Egypt are also being taken into account.
