The same technology that’s transforming your products is now transforming how the government reviews your R&D tax credits.
The IRS has been quietly deploying artificial intelligence across its audit selection, fraud detection, and taxpayer services operations, and the implications for companies claiming the federal R&D tax credit under Section 41 are significant. As of early 2026, the IRS operates 129 AI use cases, up from 54 just two years ago. For businesses claiming R&D credits, that shift changes the calculus around documentation, substantiation, and audit readiness.
Here’s a clear-eyed look at what the IRS is doing with AI, why it matters specifically for R&D filers, and what you can do about it.
The IRS Is Using AI to Choose Who Gets Audited
Audit selection at the IRS used to rely heavily on a statistical scoring model called the Discriminant Function System (DIF), which is a relatively blunt instrument that flagged returns based on deviations from population norms. That model still exists, but it’s been significantly upgraded.
AI has now been incorporated into the DIF, using an algorithm built on a large database of returns to identify those with the highest potential for inaccuracies and discrepancies between income and deductions. Returns are scored on multiple criteria, with higher scores triggering closer review. These AI systems run six times per tax year, learning with each iteration. Common triggers include year-over-year income discrepancies, extreme deduction ratios, round numbers suggesting estimates, and patterns that deviate significantly from prior filing history.
For R&D credit filers specifically, the IRS is deploying AI models across multiple taxpayer segments, shifting from broad statistical scoring to more sophisticated, relationship-driven analysis. That means your R&D claim isn’t just being reviewed in isolation; it’s being evaluated in the context of your financial history, your industry peers, and patterns the IRS has identified across thousands of similar filers.
Once Flagged, AI Assists the Examiner Too
Once a return is selected for audit, the IRS also employs AI in the examination itself. Revenue agents now have access to generative AI programs to assist with drafting information document requests and exam reports, and AI is being used to process and analyze large volumes of data in various formats to present the most relevant information to examiners.
The practical implication: Audits may move faster, but they may also go deeper. An examiner equipped with AI-assisted document analysis can surface inconsistencies across large claim packages more quickly than manual review ever allowed.
A Reduced Workforce, But the Same Enforcement Priorities
It’s worth acknowledging the current context. As of early 2026, the IRS has seen its workforce shrink by approximately 27% over the past year, and is facing further budget cuts to taxpayer services and enforcement in its Fiscal Year 2026 budget. Recent organizational changes mean fewer agents are reviewing more claims, but enforcement priorities around R&D tax credits remain unchanged. The agency continues flagging common issues including inadequate documentation, overclaimed activities, and insufficient nexus between claimed expenses and qualified research.
In other words, AI isn’t filling a gap left by reduced enforcement. Rather, it’s maintaining enforcement capacity with fewer human resources. For R&D filers, that’s a meaningful distinction.
Form 6765 Is Getting More Demanding, Not Less
The IRS’s AI push is happening alongside a major documentation overhaul for R&D credit filers. The finalized Form 6765 instructions, released in February 2026, mark a shift to a project-specific disclosure approach, moving from summary reporting to detailed business component disclosures intended to improve detection of high-risk claims at the time of filing.
The most significant change is Section G, which requires taxpayers to itemize qualified research expenses by business component, separating wages into direct research, supervision, and support categories. Section G is optional for all filers for tax years beginning before 2026, but required for tax years beginning after 2025 — with limited exceptions for qualified small businesses and filers with QREs at or below $1.5 million at the control group level.
Starting in 2026, taxpayers must report components in descending order of cost until they reach either 80% of total QREs or a cap of 50 components. For mid-market and enterprise companies with complex R&D programs, this is a material change to how claims must be prepared and documented.
Auditors will expect clear descriptions of what was tested, how it was tested, and what uncertainty was resolved, while generic summaries will not suffice, with the IRS favoring contemporaneous documentation.
What AI Cannot Do, And Where Human Expertise Still Matters
There’s an irony worth naming: The same technology the IRS is using to scrutinize R&D claims is being marketed by some vendors as a shortcut to preparing those claims. That creates real risk.
The IRS has not relaxed its substantiation standards because your documentation was generated by AI. Whether records are AI-generated or human-written, they must meet the same requirements. Auditors often request original evidence—tickets, lab reports, test results, or time records—not generated summaries. AI output may even raise scrutiny if it appears boilerplate or disconnected from the underlying facts.
R&D tax credit qualification involves judgment calls that algorithms struggle to make well. These include: Determining whether a project involved genuine technological uncertainty, whether a process of experimentation was systematic, whether activities qualify under the four-part test. These aren’t data-entry problems; they’re interpretation problems that require domain expertise.
What This Means for Your Business
If your company is claiming the federal R&D tax credit, the IRS’s AI investment raises the bar on three fronts:
Documentation quality. The IRS expects contemporaneous records proving technological uncertainty, systematic evaluation processes, and qualified expenditure tracking. Documents that were assembled retrospectively or lack specific activity-level detail will not hold up under AI-assisted scrutiny.
Business component readiness. With Section G becoming mandatory in tax year 2026 for most filers, companies that haven’t yet mapped their R&D spending to individual business components are already behind. The time to build that infrastructure is now — not when you’re preparing to file.
Audit defense. Reduced IRS staffing means extended review periods when questions arise about claims. A claim that triggers a flag but lacks complete documentation faces a longer, more difficult resolution process.
How Boast Helps
At Boast, we’ve helped more than 2,000 companies across North America access over $900 million in R&D tax credits since 2011. Our platform is purpose-built for exactly this environment: one where AI-powered scrutiny demands AI-powered preparation, backed by human expertise that understands the subjective language of R&D qualification.
Our approach combines automated data collection from your payroll, financial, and engineering systems with expert review by R&D tax specialists who know how the IRS evaluates claims. Every qualifying activity is documented contemporaneously, traceable to the underlying business component, and organized for audit defense before a notice ever arrives. That’s not a feature we added; it’s how the platform was designed.
With the IRS increasingly using AI to identify high-risk claims and new Form 6765 reporting requirements raising the documentation floor, the cost of an inadequate claim process is going up. The companies best positioned to capture their full R&D credit entitlement are those that treat substantiation as a year-round operational discipline — not a year-end exercise.
Ready to see how your current R&D claim process holds up? Get a free R&D credit assessment from Boast.
