The development of a software to analyze phone conversations to detect and prevent phone scams announced last week faces several major roadblocks, including accuracy rate and poor user knowledge of how to use the tool, say observers.
Last week, Fujitsu announced that it was collaborating with Japan's Nagoya University to develop a technology that can analyze phone conversations and detect "over trust" situations. Such situations refer to when an individual is overwhelmed with distressing information and may not have the capacity to objectively evaluate whether they are being lied to or not, such as during a remittance-solicitation fraud, it stated.
The scam detection tool functions by examining the frequency of the user's voice, as well as word-spotting voice recognition technology, Naoshi Matsuo, Fujitsu's head of technology, told ZDNet Asia in a follow-up e-mail interview. The voice recognition feature, for instance, can isolate spoken words and match it against a pre-determined vocabulary list while ignoring the rest of the conversation, he added.
The technology also analyzes the user's voice and not that of the scammer. "[As such], even if the scammer is able to act and behave naturally, this would have little effect on the overall precision of the detection of a possible phone phishing scam," he said.
Fujitsu and Nagoya University are partnering Japan's National Police Academy and the Bank of Nagoya to test the software currently. Matsuo explained that the use of technology for verification testing will give the company greater visibility on its expansion into other practical applications.
"The technology is still in its infancy," he said. "The focus now is to improve on its [current state]."
consultant, ROA Holdings
"Tech retards", inaccuracy major roadblocks
However, Yuta Torisu, consultant at ROA Holdings, pointed out that the Japanese company would face challenges in bringing this technology to market.
He noted that most frauds victims were "technology retards", and the company would have a major task on its hands to educate consumers on how to use the scam detection tool and persuade them to buy it.
"These victims are usually the elderly who are not aware of technology advancement and latest security threats, so they may not see the need for this technology and be willing to pay for it," he explained.
Tommy Tan, consultant at Tommy Tan Psychiatrist Clinic, added that while it's possible to design a tool to detect phone scammers, the accuracy of the software may be an issue.
He explained that scammers can be trained to avert detection once they understand how it works, such as choosing words that are not in the technology's pre-determined word list or coming up with methods of persuasion that do not inflict psychological stress on their victims.
Tan also said that the "over trust" technology takes about one minute to detect irregularities, which is different from lie detectors that analyze fluctuations and can give immediate feedback. "There is the possibility that the victim would have been completely scammed within that time," he warned.
Phil Hassey, owner of research firm CapioIT, reiterated that people are "naive by nature", particularly victims that are less technologically savvy. Even with user education, scams might still take place, he said, citing the success of "Nigerian scams", or e-mail frauds, that are still successfully undertaken despite widespread awareness and education.
To mitigate this challenge, Torisu suggested that Fujitsu should work with phonemakers to implement the software in their devices, which will help increase its reach and enable users to understand how it works.
When approached, Samsung and Nokia declined to comment on their interest to implement such technology on their handsets.