Last week, Interpol held a final project review of its speaker identification system, a four-year, 10 million euro project that has recently come to completion. The Speaker Identification Integrated Project, what they call SiiP, marks a major development in the international expansion of voice biometrics for law enforcement uses — and raises red flags when it comes to privacy.

Speaker identification works by taking samples of a known voice, capturing its unique and behavioral features, and then turning these features into an algorithmic template that’s known as a voice print or voice model. With enough voice prints and samples collected in its global audio database, Interpol’s speaker identification system will be able to upload an unknown voice and, regardless of the language it is speaking, match it to a list of likely candidates. SiiP’s database allow uploads and downloads of samples from 192 law enforcement agencies across the world.

SiiP will join Interpol’s existing fingerprint and face databases, and its key advantage will be to facilitate a quick identification process — say, of a kidnapper making a phone call — even in the absence of other identifiers. The platform also boasts the ability to filter voice samples by gender, age, language, and accent. When the audio recordings are taken from similar acoustical environments, accuracy rates can be extremely high.

Speech recognition technologies can identify and tag individuals every time they open their mouths, effectively ending anonymity. As Interpol’s promotional video explains, departments using SiiP can upload intercepted phone calls and also search against voices on social media. SiiP’s database will include samples from YouTube, Facebook, publicly recorded conversations, voice-over-internet-protocol recordings, and other sources where individuals might not realize that their voices are being turned into biometric voice prints. “People choose to upload material online for various reasons, but I doubt it’s to let police and arms companies then enroll them into secret databases made available to police around the world,” explained Edin Omanovic, a surveillance expert at Privacy International.

Cynthia Wong, a researcher at Human Rights Watch, warned that a broad mandate could lead to an ever-expanding collection. “There are many instances where we might consent to our voice being recorded for one purpose, but would object to using it for others, including using our voice to build and train a massive voice biometric database and recognition system,” she said. “Or perhaps we didn’t consent to our voice being recorded at all — perhaps our voice was secretly recorded or inadvertently caught in the background of a recording, but has now been placed in Interpol’s database.

SiiP’s database will include samples from YouTube, Facebook, publicly recorded conversations, and other sources where individuals might not realize that their voices are being turned into biometric voice prints.

Interpol’s system appears to represent the largest international voice-harvesting effort to date. But as a recent Interpol survey reports, many law enforcement agencies have already been quietly using voice recognition systems. More than half of the 91 departments in 69 countries surveyed already run automated speaker identification programs. In 2010, Mexico announced that it had created the world’s first nationwide automated speech recognition system, using Russia’s Speech Technology Center. As The Intercept reported in January, the National Security Agency has been using speaker recognition to monitor terrorists, politicians, drug lords, spies, and agency employees since at least 2004. The Department of Homeland Security’s new HART database — which already stores a fraction of the world population’s biometrics — will support voice data, in addition to DNA, scars, and tattoos. Human Rights Watch released a bombshell report last year exposing China’s state-of-the-art system, which appears to be integrated with mobile phone networks and recognizes the voices of known suspects in real time. Wong says that China’s integrated surveillance system is likely to set a precedent for other countries to follow suit.

It is unclear what kinds of legal safeguards exist to protect against the creation of international biometric collections, which may retain data indefinitely. Oversight bodies for intelligence sharing are scarce. “How can Interpol ensure that the voice samples other agencies submit are lawfully intercepted?” Wong asked. “Based on our work, many countries around the world fall woefully short in regulating surveillance to begin with.”

Wong, Omanovic, and others are also concerned about the plans to create a “blacklist” of criminal and suspect voices. It is not known who will audit the list, what rules will govern when and how countries add to it, and what happens if someone is placed on the blacklist in error, Omanovic said. Wong wonders what kinds of safeguards will be used in countries where the government regularly criminalizes dissent or jails journalists.

An Interpol report researching the ethics of biometrics acknowledges the great chilling effect that such collection might have. “Processing of audio data that is never actually viewed does not count as intrusive in itself, though it can lead to future intrusion,” it states. “Particularly in an investigatory context where the aim is identifying suspects for future scrutiny. Even when no human viewing does take place, the application of these techniques has consequences resembling genuine intrusion: They can easily cause people to fear that their information will be exposed, creating a harmful incentive to avoid associational life or unconventional behavior.”

Verint, a multinational biometrics company, has been leading project development, and did not respond to a request for comment. As Privacy International has documented, the company has been known to install and “sell intrusive and mass surveillance systems worldwide including to authoritarian governments.” This, according to Omanovic, “tells you everything you need to know about the level of ethical scrutiny being applied. The EU is being securitized without any public debate, and the only people who will ultimately benefit are those working in the arms industry.”