The Department of Homeland Security approved thousands of noncitizens to work in the U.S. even though its own system tried to flag them as probably ineligible, according to a new inspector general report that paints a grim picture of the government’s best tool for weeding out undocumented immigrant workers.
The E-Verify system also confirmed work authorization for about 280,000 noncitizens in 2019 without actually matching them to photos online, opening an avenue for fraud that undocumented immigrants could exploit to take jobs, the Homeland Security inspector general said.
Over the years, the biggest concern about E-Verify was that it would wrongly reject citizens or immigrants who were permitted to work. But the new report suggests the larger issue is that U.S. Citizenship and Immigration Services is approving people who should not be taking American jobs.
“Until USCIS addresses E-Verify’s deficiencies, it cannot ensure the system provides accurate employment eligibility results,” the inspector general said in the report, released Wednesday.
That’s a fairly blistering critique of a system that had been held out as the key tool for reducing illegal immigration in any future immigration deal reached on Capitol Hill.
Under E-Verify, employers run information from new hires through the system and get back either a confirmation of work eligibility or a tentative nonconfirmation. In those cases, a more full review is done to determine whether the hire is indeed eligible.
In fiscal 2019, nearly 40 million E-Verify checks were performed. In 98% of those, the worker was quickly authorized. Another 1.6% got tentative nonconfirmations, while less than 0.4% were given final nonconfirmations indicating they could not be hired.
The system works by querying a series of government databases to gauge work authorization.
If it cannot locate the right documents, it estimates a “confidence score” from 0 to 100 gauging how likely it is that the person is work authorized. Those with a 50% level of certainty or above are confirmed. Originally the level was supposed to be 70%, but the agency figured that would create too much additional work to review the extra cases that fell in that range.
According to the inspector general, nearly 3,000 applications — about 1% of the cases that were given a confidence score check — were approved during the period under study, despite falling below the 50% level.
In about 2,700 of those cases, the confidence score was a flat zero.
“The error went undetected for 6 months, from September 2018 to March 2019,” the auditors wrote. “According to USCIS officials, a system update to E-Verify or a data sharing error with one of its partner systems may have caused the issue. USCIS officials could not determine the exact error that led E-Verify to bypass the 50% confidence score requirement and, therefore, could not prevent it from reoccurring.”
And another 4,000 cases were approved in 2019 even though the agency didn’t fully match their visas with their employers’ situations.
USCIS knew of this problem but said it wasn’t big enough to fix, given the tradeoffs. The agency said the alternative was to give more people initial rejection notices, even though they were eligible, then have them follow up to clear up their cases.
But the auditors said USCIS never did any formal analysis to back up that claim.
Tracy Renaud, who was acting director at USCIS at the time the audit was finalized last month, said the agency has taken steps to improve E-Verify, including monthly audits to try to spot problems like the confidence score confirmations snafu.
She bristled, though, at some of the criticisms, such as the troubles with automated photo-matching. She said E-Verify isn’t required to have photo-matching, and in some cases it’s impossible. For example, some states refuse to share their driver’s license data, which is a key source of photos.
Other times, state databases are offline, she said.
Still, the agency concurred in all 10 of the inspector general’s recommendations for improvements.
• Stephen Dinan can be reached at sdinan@washingtontimes.com.
Please read our comment policy before commenting.