New Delhi: Voice samples from “military sensitive areas of India”, including Jammu and Kashmir and Punjab, were collected by an AI firm based in Beijing via Indian intermediaries and sent to Chinese agencies for “use and analysis.” is sold to Based think tank claims.
New Kite Data Labs, which explores how China uses and leverages data, found in a report that Beijing-based company Speechocean has close ties to China’s security services and the People’s Liberation Army. claims to have
It is suspected that data collected by this company could be used by China to engage in “automated extraterritorial mass surveillance”.
New Kite Labs founder and academic Christopher Balding told ThePrint that Speechocean (SO) is working with a New Delhi-based subcontractor, a Business Process Outsourcing (BPO) firm, to provide Said they were looking for individuals to record their voices. A region of India, especially a militarized region.
“These people are paid a small fee to record phrases, words and sentences in their own language and accent. These recordings are collected using the Speechocean app that can be downloaded to their phones. So people from Kashmir, Punjab were identified and paid to record audio samples without revealing the purpose.These samples were then sold to China,” the investigation said. argued Balding, who led the
Regarding the impact of voice data sent from India to China, Boulding claimed that Speechocean was “known” to be sold to the Chinese military.
“Speechocean’s attempt to obfuscate their activities on behalf of Chinese security agencies raises legitimate security concerns, as this data is used to train technological tools involved in mass surveillance outside of China. It implies that it is used in ,” says Balding.
Speechocean describes itself on its website as an artificial intelligence data resource provider dedicated to providing “engineering data products and services to enterprises and scientific research institutes across the AI industry chain.”
Balding told ThePrint that New Kite Labs has notified the Indian security services of its findings.
The information is under investigation, sources with the Indian security services confirmed.
read also: China is gearing up for a full-spectrum AI war.India is still 15 years behind
“Absolute proof that SO worked in Kashmir, Punjab”
A Chinese database with a large number of Indian IP addresses led New Kite Labs researchers to Speechocean. Speechocean is a Shanghai Stock Exchange-listed data provider that produces datasets for training and developing algorithmic models.
Balding said SO uses local intermediaries to collect voice data from India, especially sensitive areas.
“SOs are active in Punjab and Kashmir and there is absolute evidence of that at every level,” Balding said.
“We obtained log files of voice file transfers sent from Indian IP addresses in Punjab and Kashmir to a Speechocean database in China. It goes all the way back to recruitment efforts,” he added.
Balding told ThePrint he was “worried” because of the company’s apparent ties to the People’s Liberation Army and other Chinese security agencies.
“The company is known to sell to the PLA Cyber Warfare Division. There are documents stating that Speechocean was bidding to sell Vietnamese data to the PLA Cyber Warfare Division. is SpeechOcean’s primary business model: they collect and sell data,” he argued.
When asked about the nature of the voice sample taken in India, Balding said it was “unknown.”
“We are not clear on the nature of the voice samples coming from India to China as we do not have access to the raw files,” he said.
ThePrint reached out to Speechocean by email, but received no response at the time this report was published.
Links to Chinese military and security agencies
According to the website, Speechocean was founded in 2005 by the current chairman, He Lin.
According to reports, as of September 2021, he was with Cai Huizhi, founder and chairman of Beijing Zhongke Cruise Digital Technology, a listed Chinese defense company that provides key submarine-related technology to the Chinese military. was married.
“The company’s website includes a video of President Xi Jinping inspecting a military facility equipped with the company’s technology. It highlights the scope,” the report said.
The SO is “deeply embedded in the national security apparatus,” because China’s National Computer Network and Information Center is the company’s non-founder-funded shareholder, investor, and customer, the report argues. I’m here.
The government agency “is responsible for China’s Internet security and censorship and invests in SO through investment funds and holding companies,” the report continues.
“Their mission is to localize technological development and progress to make China a global technology leader and help promote and defend national security in information management,” it added.
The think-tank also claimed that it had documents showing it was not only involved in maintaining public security within China, but also “cooperating with security intelligence agencies targeting foreign countries.”
“We have identified an open tender for a project related to Vietnamese speech classification for the People’s Liberation Army Strategic Support Force (SSF), better known as the Cyber Warfare Division. It is related to a classification project related to the project and a classification project for Chinese minority languages in northwestern China,” the report said.
Data trace back to Beijing, Hong Kong
According to a New Kite Labs data report, the data collected by SO included voice samples containing words, phrases, or dialogue of specific “accents and nationalities” and were recorded in three different locations: Beijing, Hong Kong, and Germany. We were able to trace it back to the main IP address. .
“Data was tracked to servers in Frankfurt, Germany registered with Aliyun Computing in Beijing, Alicloud in Hong Kong, and Alibaba Singapore,” the report said.
Beyond its ability to collect and store data, the report also notes that China has “developed technological capabilities through software for automating behavioral surveillance techniques that are typically assisted by AI (artificial intelligence)/ML (machine learning) applications. It said it has deployed a “primary resource for creating”.
(edited by Asavari Singh)
read also: Government tells agencies to coordinate crackdown as Chinese firms also have ‘security concerns’
Comments
Post a Comment