The data labeling startup Scale AI Inc. is being investigated by the U.S. Department of Labor, amid suspicions that it’s violating the Fair Labor Standards Act, the company said today.

The Labor probe is investigating Scale AI’s compliance with fair pay standards and working conditions described by the act. It’s believed that it was launched almost one year ago during previous U.S. President Joe Biden’s administration, but it has only just now come to light.

According to the startup, it has been working closely with the department to explain its business model and how things work in the broader artificial intelligence industry. The Fair Labor Standards Act is a federal law that covers unpaid wages and illegal retaliation against workers, and it also defines who should be classified as an employee or a contractor.

Just because Scale AI is being investigated doesn’t necessarily mean that the company is guilty of wrongdoing, and it may be that no charges are brought against it.

Founded in 2016, Scale AI is a provider of AI training datasets. It supplies AI model developers with the vast amounts of accurately labeled data they need to train sophisticated large language models, such as OpenAI’s GPT-4.

The company does produce its own datasets, but most of them are created by third-party contributors, who also label the information within them. In this way, it’s also a gig work platform for researchers and data scientists, who create and fine-tune customized datasets on request.

Scale AI claims to have tens of thousands of contributors working for it in more than 9,000 cities and towns in the U.S.

A spokesperson for the company told TechCrunch that it strongly disputes allegations that it’s underpaying or mistreating its contributors. He added that the company is in full compliance with the Fair Labor Standards Act, and strives to ensure its pay rates provide contributors with a living wage, based on local standards. He further insisted that almost all of its payments are made on time, and that more than 90% of inquiries regarding payments are resolved within three days.

The spokesperson complained that regulators misunderstood its work in building, testing and evaluating AI when they launched their investigation last year. He stressed that the company provides “flexible work opportunities” to Americans, and that it has had overwhelmingly positive feedback from contributors.

On its website, the Department of Labor says the vast majority of cases it deals with are resolved administratively, but it does have the power to prosecute employers who violate the law, with hefty fines and possible imprisonment for the worst offenders. It also has the ability to force companies to reclassify contractors as employees.

This isn’t the first time Scale AI has come under scrutiny for its labor practices. In 2023, the Washington Post reported that overseas contributors had complained of low pay and impossible demands. The company responded back then, saying that its rates of remuneration were constantly improving. More recently, in December and January, two former workers filed lawsuits against the company, alleging they were underpaid and misclassified as contractors rather than employees, which meant they were denied overtime pay and sick days.

Scale AI has much stronger ties to the new White House administration under President Donald Trump. Founder and Chief Executive Alexandr Wang was one of several tech CEOs who attended Trump’s inauguration in January, and former Managing Director Michael Kratsios previously served as the country’s chief technology officer during Trump’s first term as president.

Kratsios has since been nominated by Trump to serve as the director of the White House’s Office of Science and Technology Policy, where he will advise the president on matters relating to science and technology.

The company’s ties with the government appear to be advantageous for the company. Earlier this week, Scale AI announced it had signed a multimillion-dollar deal with the U.S. Department of Defense, under which it will help it to train AI agents for military planning and operations.

Image: Scale AI