A majority of Americans trust law enforcement to use facial recognition technology responsibly but are less comfortable with private sector usage, according to a new survey
A new research reveals how Chinese facial recognition terminals are easily fooled by a tester wearing a 3D mask. AFP / DAVID MCNEW

A majority of Americans trust law enforcement to use facial recognition technology responsibly but fewer are comfortable about its deployment by the private sector, a poll showed Thursday.

The Pew Research Center survey found US adults have confidence in law enforcement on using the artificial intelligence systems by a 56-39 percent margin, and a larger majority endorse the use of the technology to assess security threats in public spaces.

But trust levels are markedly lower for private entities using facial recognition.

Only one in three said they trust technology firms and just 18 percent said they believed advertisers would use facial recognition responsibly.

Just 36 percent said it would be acceptable for these tools to track who is entering or leaving apartment buildings; 30 percent approved the monitoring of attendance of employees at a place of business; and 15 percent endorsed the idea of seeing how people respond to public advertising displays in real time.

The results varied by age, political affiliation and racial or ethnic background.

Roughly three-fifths of whites but only 43 percent of blacks said they trust law enforcement to use facial recognition responsibly. Republicans were more trusting than Democrats and older adults more than their younger counterparts.

Among US adults, 86 percent have heard at least something about facial recognition technology and 13 percent said they have not heard anything about it.

The poll comes amid growing concerns about facial recognition by law enforcement and the private sector. San Francisco has banned the use of the technology by official agencies, and some researchers have warned of errors, notably in identifying minorities, and of the creation of large databases which could be breached or hacked.

The report is based on a survey of 4,272 US adults from June 3-17, with an estimated margin of error of 1.9 percentage points.