Sometimes tax compliance robots follow the law, but sometimes they break it.
In 2020, as governments respond to COVID-19, new laws and changes to existing laws are emerging with remarkable speed. Governments are dialing up some regulatory obligations while dialing other regulations down. It requires a lot of candle power—or computing power—to keep up with the increasing complexity.
What enables legal complexity? In part, private computing power. TurboTax and other tax software programs have made unbearably complex tax laws administrable. Private automated systems—or compliance robots—also help administer the increasing legal complexity that results from changing legal rules such as those resulting from the response to COVID-19.
Private automated systems played a role from the beginning of the pandemic response. Alphabet’s affiliate Verily developed methods to screen and test for coronavirus. Apple and Google partnered to develop contact-tracing software for smartphones. Kabbage helped small businesses apply for the federal government’s Paycheck Protection Program. And TurboTax allowed individuals to claim their $1,200 economic impact payments.
Do such compliance robots actually help promote compliance? Do these robots themselves follow the law?
Tax law has some experience with this question. I analyzed that experience in a recent article published in the Ohio State Technology Law Journal. Available information suggests that algorithmic tax compliance robots, such as TurboTax or H&R Block Online, interpret substantive tax law conservatively. They appear to follow Internal Revenue Service (IRS) guidance, even though the guidance may oversimplify the underlying law. Tax compliance robots apparently aim to minimize audit risk.
This conservative posture can burden some taxpayers by limiting what they can claim. For instance, my testing suggests that tax compliance robots do not offer taxpayers different options for reporting income from the sale of blood plasma. The Taxpayer Advocate Service has reported that some automated compliance systems have omitted information about how to claim some casualty loss benefits.
On the other hand, tax compliance robots do not systematically encourage taxpayer honesty. True, they reproduce the required “under penalties of perjury” certification at the end of tax returns—but tax software programs do not contain other tools to discourage fraud. They do not use suggestions such as requiring attestations of honesty at salient moments or providing pre-filled tax returns.
Instead, users buy a product that allows the overstating of deductions or the understating of income to achieve a lower tax bill or a bigger refund. Tax software includes prominent displays of the dollar amount of the tax or refund due. The amount adjusts continually with new data inputs and may encourage users to lie to reduce their taxes.
In addition to the questions of substantive tax compliance and measures to deter taxpayer fraud, there is the question of how tax compliance robots use taxpayer data.
Under a Free File Agreement between the IRS and the Free File Alliance—a group of tax software companies including Intuit and H&R Block—the IRS agrees not to develop free-filing software of its own. In exchange, the companies agree to provide free filing services for taxpayers at or below a certain income threshold, which is about $69,000 in 2020. These companies also agree, for Free File returns, to keep taxpayer information confidential and to direct users to alternative options if their free tax software does not support a particular return.
But it appears that these firms and their tax compliance robots have systematically violated the taxpayer confidentiality and data provisions of the Free File Agreement. The companies have required users to allow the firms to use the information for cross-selling and other purposes. They have also engaged in upselling rather than directing users to other free alternatives.
These apparent breaches are the subject of several lawsuits filed in 2019 alleging, among other claims, violations of California’s unfair competition statute. Whether these developments change how tax compliance robots use taxpayer data, at least for Free File returns, remains to be seen.
Why do tax compliance robots sometimes follow the law and sometimes break the law? One might think that they would follow the law when they are liable for illegality. But that is not how it works.
Tax compliance robots appear to follow substantive tax law even though they are not liable for breaking it. Tax law provisions generally would not hold compliance robots liable for bad or aggressive advice. Tax compliance robots would not face meaningful liability under private law either, despite so-called “audit protection provisions” in user contracts.
But the area where they do face a direct requirement relates to taxpayer confidentiality and data law obligations for taxpayer returns covered by the Free File Agreement. Yet in this category, they appear to break the law.
Instead of liability driving whether tax compliance robots follow the law, market incentives align with whether tax compliance robots follow the law. Intuit and H&R Block explain that “peace of mind” is part of what they are selling to taxpayers who use their software. That translates to conservative substantive tax law interpretations that follow IRS guidance.
Tax compliance robots are not, though, selling a taxpayer honor code. It apparently does not pay for a tax compliance robot to resist user fraud actively.
With respect to confidentiality and data limitations, if tax compliance robots chose to follow the law and refrain from exploiting Free File taxpayer data, this decision would strike at the heart of their business model. H&R Block has acknowledged, for instance, that its business objectives include “the migration of clients from its free tax service offerings to those for which it receives fees.” Intuit has warned that “if it is not able to effectively…utilize its customers’ data, its revenues may be harmed.” It would be painfully costly for firms to obey data protection and taxpayer confidentiality law. It is not surprising that, historically, they have not done so.
Automated private sector compliance systems may make modern government possible, especially during an unprecedented pandemic. As law becomes more complex, compliance robots will find themselves increasingly at the center of public administration. But private automated systems do not answer only to the public interest. They also answer to the market. Sometimes compliance robots will follow the law. Sometimes they will not.
With increasingly complex modern regulation comes the task of monitoring compliance robots’ decisions and holding them accountable when the market does not. If we want compliance robots to guard against fraud, or to keep user data confidential, then there ought to be a law that requires these results and holds firms liable when automated systems fail to uphold the law’s requirements. One model is the state law-based unfair competition lawsuits that target firms’ upselling practices for Free File returns, which should help bring an end to this practice.
Greater regulation of automated systems that implement substantive legal guidance might itself increase legal complexity. But compliance robots have years of experience with layers of intricate tax law provisions. They can handle it.