A Remote Developer You Treat Like Staff Is Legally an Employee, Not a Contractor

Key Takeaways

Misclassifying a contractor as an employee creates real financial exposure - back taxes, employer FICA, wage claims - not just a paperwork problem. Three different tests decide the question: the IRS common-law test, state ABC tests (California plus 25+ other states), and the DOL's economic-reality test under the FLSA. They can disagree on identical facts.

The federal DOL test is currently unsettled: the 2024 six-factor rule is still cited in private litigation, but DOL proposed rescinding it in February 2026 for a 2021-style test weighting control more heavily.

Audit triggers are behavioral, not contractual: fixed hours, exclusive availability, company-issued tools and accounts, and integration into core engineering work all read as employee status, whatever the contract says.

A signed contract alone doesn't fix misclassification. Autonomy, invoicing structure, and a documented pattern of independence do the legal work the contract only describes.

When you want a compliant counterparty for the contractor relationship itself, rather than managing classification risk in-house, routing it through a Contractor-of-Record is the structural fix.

The "we just hire contractors, we're fine" trap

The contract you sign with a contractor has almost no bearing on whether a tax agency, a state labor board, or a court later agrees that the person is legally a contractor at all. That's the trap a lot of small engineering teams walk into when they start hiring remote contractor-developers across borders: get a standard independent-contractor agreement signed, treat classification as a closed question, and go back to shipping the product.

Picture a fairly typical setup. A handful of remote backend and frontend developers, spread across two or three countries, working inside the same Slack workspace and Jira board as the in-house team, with seats in the same GitHub org and sometimes the same SSO. Each one signed a one-page contractor agreement on day one. Nobody has reopened it since.

The working assumption is that the signed agreement is the compliance work, and once it's filed away, the classification question is settled. It isn't, and it's worth being precise about what this question even is, since "contractor compliance" gets used for two unrelated things. This isn't about jobsite safety, insurance certificates, or vendor prequalification software. It's worker classification law: who counts, legally, as an employee versus an independent contractor, and what happens to a company when it gets that call wrong.

Employer-level misclassification rates run an estimated 10-30%, and at least a dozen US states proposed or passed misclassification-related legislation in 2025-2026 alone, mostly to give enforcement more teeth. The contract sitting in a shared drive isn't what a regulator looks at first.

What misclassification actually is, and why it bites

Misclassification is treating a worker who legally meets the tests for employee status as an independent contractor instead, usually to skip payroll tax withholding, benefits, and the rest of the employment-law stack that comes with a W-2. Intent doesn't determine whether it happened. A team that genuinely believed its remote contractor-developers were correctly classified can end up owing close to the same back taxes and penalties as one that classified them as contractors specifically to cut costs. Intent does change the penalty math, though, which is the useful part covered below.

Getting the classification wrong opens up four categories of exposure, and they stack rather than substitute for each other:

• Back payroll taxes plus the employer's own FICA share, reaching back as far as the IRS or a state agency can audit.
• Wage-and-hour claims under the FLSA and state law, if the worker should have been paid overtime or minimum wage as a non-exempt employee.
• Retroactive benefits and ERISA exposure, where a reclassified worker would have qualified for the company's existing benefit plans.
• State unemployment-insurance and workers'-compensation liability, which run through state agencies on their own timeline, independent of anything the IRS decides.

Cost math: section 3509(a) vs. 3509(b) vs. VCSP on five reclassified developers

Take five remote contractor-developers reclassified as employees, each earning $120k a year - $600k in total wages. 26 U.S.C. §3509 gives an employer two different discounted rates for unintentional misclassification, depending on whether 1099s were actually filed for those workers:

Run your own headcount and comp numbers through the same formula rather than trusting a single scary total - the gap between filing 1099s and not filing them alone doubles two of the three cost components.

One more wrinkle: none of this discount applies if the failure to withhold was intentional. Section 3509 relief is unavailable where the failure to withhold was due to intentional disregard of the withholding requirement, which pushes an employer back to the full, undiscounted liability. Willful failure to collect and pay over employment taxes is also a separate felony under IRC §7202, carrying up to $10,000 in fines and five years in prison.

How regulators and courts actually decide contractor vs. employee

Three separate tests decide worker classification in the US, and they don't share one definition of "contractor": the IRS common-law test, state-level ABC tests, and the federal DOL rule under the FLSA. A team can pass one of these and fail another on the exact same fact pattern. That's the practical reason a signed contract doesn't settle anything on its own - none of the three tests asks what the contract says first.

The IRS common-law test: behavioral control, financial control, relationship of the parties

The IRS groups evidence of control and independence into three categories, per its guidance on employee vs. independent contractor status: behavioral control (who directs how, when, and with what tools the work gets done), financial control (who bears the risk of profit or loss, whether the worker has unreimbursed expenses or a real investment in their own equipment), and relationship of the parties (written contracts, how permanent the engagement is, whether benefits are provided). No single factor is dispositive.

The IRS's own instruction is that all evidence of control and independence has to be weighed together, and which factors matter most shifts by occupation and industry - there's no fixed scorecard. If a company or worker wants an official answer instead of a best guess, either party can file Form SS-8 and have the IRS make the determination directly, though that process runs on the IRS's timeline, not the business's.

The ABC test: used in California and 25+ other states

State-level ABC tests work differently from the IRS test: instead of weighing factors, they start from a presumption that the worker is an employee, which the hiring company has to disprove. California's ABC test requires satisfying all three prongs:

• (A) the worker is free from the company's control and direction, in the contract and in actual practice;
• (B) the work falls outside the company's usual course of business;
• (C) the worker is customarily engaged in an independently established trade of the same kind.

Fail any single prong and the worker counts as an employee for state-law purposes - unemployment insurance, wage law, workers' comp.

Prong B is where dev teams get caught. A software company that hires a contractor to write its own product code is asking that contractor to do the exact thing the company exists to do, and California and Massachusetts read "usual course of business" narrowly enough that this sits close to a straight fail rather than a gray area a well-drafted contract can argue around.

ABC tests apply based on where the worker is located, not where the hiring company is headquartered, so a US company with a contractor working from California is subject to California's version regardless of the company's own state.

The federal DOL rule, and why it's currently in flux

As of this writing (checked July 2026), the federal test under the FLSA is genuinely unsettled, not a stable fourth option to add to the list above. The Department of Labor's 2024 rule, effective March 11, 2024, set a six-factor "economic reality" test: opportunity for profit or loss, the worker's own investments, permanence of the relationship, degree of control, whether the work is integral to the business, and skill or initiative. No factor gets fixed priority over the others.

DOL announced in May 2025 that it would stop enforcing that rule, and on February 27, 2026 it published a formal proposal to rescind it (NPRM RIN 1235-AA46; comments closed April 28, 2026). The replacement on the table reverts to a 2021-style structure that elevates control and opportunity for profit or loss to "core factors," with skill, permanence, and integration carrying less weight.

None of this makes the 2024 rule dead law yet. It's still cited in pending private FLSA litigation, since courts run their own economic-reality analysis and aren't bound to follow whatever version DOL currently enforces. DOL field staff, meanwhile, aren't applying the 2024 six-factor test in practice. Treat "the DOL test" as a moving target through 2026 rather than a settled fourth checklist, and check the DOL's rulemaking page for a final rule before relying on this section for anything time-sensitive.

A worked example: running all three tests on one remote developer

Here's a hypothetical, built purely to show how the three tests interact, not a real client: a startup engages a remote backend developer at full-time-equivalent hours, working exclusively for the company for eight months with no other active clients, using the company's Slack, Jira, and GitHub organization under a company-issued account, and following a sprint schedule the engineering manager sets.

Run that fact pattern through the IRS, ABC, and DOL tests side by side, and the disagreements between them get concrete fast.

The tests converge on the two loudest facts in this pattern: the company controls how and when the work happens, and the work itself is the company's core product, not something adjacent to it. That's what fails ABC's prong B and the IRS's behavioral-control factor for nearly identical reasons.

Where the tests can genuinely split is financial control: a developer who owns real equipment and operates as a separate registered business has a fact that helps under the IRS test, even though ABC's prong B fails regardless. A strong prong A or C can't rescue a failed prong B.

Run your own contractor engagements through this same table before an auditor does it for you. This exact fact pattern shows up often enough on early-stage engineering teams that it's worth checking against a live roster, not just a hypothetical.

What actually flips a contractor into an employee during an audit

Auditors and courts don't start with the contract. They start with how the relationship actually runs, then check whether the paperwork matches what they see. When the two disagree, behavior wins. A carefully worded MSA doesn't stop a reclassification finding by itself.

The list below isn't the IRS, ABC, and DOL tests explained again. These are the observable things an investigator, or a plaintiff's attorney, checks first, because each one is a fast, checkable proxy for behavioral or financial control:

• Fixed or company-set hours. A schedule the manager assigns reads nothing like a contractor working whenever they choose.
• Exclusive availability. A contractor with no other clients looks like a de facto employee.
• Company-issued tools and accounts. A laptop, Slack workspace, GitHub org seat, or SSO login the company provisions and controls.
• Integration into core engineering work. Writing the product's main codebase, attending sprint planning, and participating in on-call rotations.
• No meaningful profit-or-loss risk. Flat hourly or monthly billing with no bidding, no expenses the contractor bears, and no upside from efficiency.
• Indefinite duration. Eight months, twelve months, or ongoing with no end date reads as permanence, not a project engagement.
• Managerial direction on how to do the work, not just what to deliver. Code review feedback, assigned tickets, and mandated tooling all count.