NSA voice recognition capabilities appear to be more advanced than previously thought
The National Security Agency has had voice recognition technology since 2006, and developed capacities to identify previous recordings of individuals using unknown phone numbers, secret codes, or multiple languages, according to classified documents examined by the Intercept.
The NSA’s voice recognition technology was used during Operation Iraqi Freedom to verify deposed leader Saddam Hussein as the voice in recorded media, and the Intercept traces NSA voice recognition use back to at least 2003. In addition to alleged spies and foreign targets, the documents suggest that the NSA planned to use the technology to prevent whistleblowers like Snowden.
“Our voice is traveling across all sorts of communication channels where we’re not there,” Timothy Edgar, a former White House adviser to the Director of National Intelligence told the Intercept. “In an age of mass surveillance, this kind of capability has profound implications for all of our privacy.”
Trevor Timm, executive director of the Freedom of the Press Foundation, points out the capability the technology provides to identify anonymous sources, and the additional risk to privacy caused by the ubiquity of microphones embedded in devices such as smartphones.
A combination of NSA voice recognition technologies allows analysts to search hundreds of hours of audio recordings in seconds looking for keywords or a certain individual. The article outlines a program called Voice in Real Time or Voice RT, which provided a “one-stop shop” for translation, identification, transcription, and search, which was shared with the UK’s GCHQ, and another called HLT Lite, which identifies modified or anomalous voices.
Finnish researchers published a study in November showing that advanced voice recognition systems could be duped by skilled voice impersonators, but the findings may not apply to HLT Lite.
The article also outlines a number of legal, social, and privacy concerns raised by the various systems and their potential applications.
In response to questions from the Intercept, the NSA said: “In accordance with longstanding policy, NSA will neither confirm nor deny the accuracy of the purported U.S. government information referenced in the article.”
Albert Gidari, director of privacy at the Stanford Center for Internet and Society, told the Verge that the technology creates an inherent problem with storage of data which can be biometrically analyzed, such as voice identification databases created by voice assistants like Google Home and Amazon Echo. The companies have not made clear how much data government agencies are requesting or receiving specifically from voice assistants.
As previously reported, Human Rights Watch alleged in October that the Chinese government has been collecting voice patterns from tens of thousands of people to monitor citizens for counterterrorism and “stability maintenance” purposes.