A San Francisco AI startup focused on identity verification has set lofty goals to secure a government contract.
Checker CEO Daniel Janisse told Business Insider that the company hopes to help governments reduce “fraud and waste” by not only screening new employees, but also verifying eligibility for benefits such as Medicare and Social Security.
Janisse said the company isn’t ready to launch a product yet, but said a frictionless government-backed system may be years away.
But AI and safety experts told Business Insider there are legal and technical hurdles for any company to take on the challenge of leveraging AI to automate benefits systems.
Checkr primarily uses AI to perform background checks and surface information such as criminal records and motor vehicle reports. The company has major deals with Uber and Lyft to curate new drivers, and is valued at more than $5.7 billion after raising $120 million in funding in 2022. In 2025, Checker reported more than $800 million in revenue and had more than 120,000 customers.
Checkr asks what he wants. “There’s a lot of fraud going on and bad actors are just getting government money instead of the right people who need it,” Yanisse said of Medicare and other programs, adding that it’s very difficult for the government to actually verify people’s employment status and income.
The Medicare Fee-for-Service program estimates that there were $28.83 billion in “improper payments” at a rate of 6.55% in 2025, but not all such cases are the result of intentional fraud. Payments to individuals who do not provide sufficient documentation and whose income level is not verified are also considered inappropriate by Medicaid.
“Unfortunately, with the advent of AI, there will be more fraud, impersonation, and fraud,” Janisse said. “There’s a lot of friction, a lot of repetition, and now there are deepfakes as well.”
A Checkr spokesperson told Business Insider that the company’s potential involvement with the government is “still conceptual at this time.”
The company also pointed to an investigation by corporate identity verification platform Middesk that found that of the $1.09 trillion in Medicaid payments made to approximately 1.6 million health care providers between 2018 and 2024, $563 million in payments went to health care providers blacklisted by federal health care programs for criminal activity or fraud.
Automating identity verification can be difficult
Stuart Russell, a computer science professor at the University of California, Berkeley and an AI pioneer, told Business Insider he is “not optimistic” that plans to use AI to determine eligibility for benefits will work as advertised.
“Some versions of this type of AI system, or LLM, are unable to generate a valid explanation for their decisions, making it impossible to challenge incorrect decisions,” Russell said.
Russell also cited the European Union’s General Data Protection Regulation, which prohibits decisions that have significant legal implications for individuals from being made by fully automated systems.
Baobao Zhang, the Maxwell Dean at Syracuse University and an associate professor who specializes in AI politics, told Business Insider that while it’s impossible to accurately assess how good Checker’s verification system is right now, past attempts by governments to combine people’s interests with automated systems are alarming.
“If the federal government or any other state government is looking to contract with a vendor to automate welfare fraud detection, history has shown that the risks are high and require serious real-world evaluation before deployment,” Zhang said.
In Indiana, an attempt to streamline and automate the welfare eligibility system by outsourcing contracts to IBM ended in a legal battle, with the state suing the company for $1.3 billion for the canceled project in 2010. Based on court records, the Indiana Department of Family and Social Services said processing errors from IBM led to incomplete benefit denials and harmed people in need.
In Australia, an automated government program called Robodebt designed to detect fraud used a flawed algorithm to instruct welfare recipients to repay their benefits, sending them letters claiming they owed thousands of dollars. Australia’s top public inquiry body, the Royal Commission, has revealed that at least three people died by suicide after being falsely instructed to repay debts they did not owe from robo-borrowers. The system was ruled illegal by a court in 2019.
Ifeoma Ajunwa, founding director of Emory University’s AI and Future of Work Program, told Business Insider that when a government agency implements AI, it should create an advisory board made up of engineers and social scientists to give a voice to affected constituents.
“I think we need to tread carefully when delegating government functions to AI technology,” Ajungwa said. “While these tools are touted to increase efficiency and reduce costs, we also need to establish guardrails for using them to protect the public.”
