Fact-Check Any Claim. Instantly.

Real sources. Independent analysis. Human review.

Claim analyzed

“Major software companies report that the vast majority of their source code is now written by AI.”

The Conclusion

The claim is
False
3/10

Executive Summary

The claim is false. Current evidence shows only 30-42% of code at major software companies is AI-assisted, not the "vast majority" claimed. Microsoft reports ~30%, industry surveys show 42% maximum—all well below majority thresholds.

Warnings

  • The claim conflates 'AI-assisted' code (human-written with AI suggestions) with 'written by AI' (fully AI-generated), creating misleading impressions of AI autonomy
  • Current figures of 30-42% are projected to grow to 65% by 2027, but the claim asserts companies currently report vast majorities when they do not
  • Some sources are unreliable marketing compilations or vendor reports with promotional incentives rather than transparent company disclosures
Full Analysis

The Claim

How we interpreted the user input

Intent

User wants to verify whether major software companies have publicly reported that most of their code is now AI-generated

Testable Claim

The user's input, neutralized and hardened into a testable hypothesis

“Major software companies report that the vast majority of their source code is now written by AI.”

The Research

What we found online

Summary of Findings

18 sources used 4 supporting 4 refuting 10 neutral

All sources are listed in the Sources section at the end of this report.

The Debate

The for and against arguments

P
Proponent Argues TRUE

Multiple widely cited industry reports show AI is already producing a near-majority share of code—developers estimate ~42% of committed code is AI-generated or AI-assisted today (Source 4, ShiftMag; Source 8, SonarSource), and major-company leadership has publicly quantified substantial AI code generation inside top software firms like Microsoft at ~30% (Source 10, Fortune). Given that these figures are current and trending sharply upward (Source 4, ShiftMag; Source 8, SonarSource) alongside near-ubiquitous AI tool adoption among developers (Source 5, JetBrains Blog; Source 6, Stack Overflow), it is reasonable to conclude that major software companies can and do report that the vast majority of their source code is now written by AI in practice (especially when “written by AI” is understood to include AI-generated and AI-assisted code).

O
Opponent Rebuttal

Your argument conflates "AI-assisted" with "written by AI" to artificially inflate the numbers, but even your own sources contradict the claim—Source 10 (Fortune) explicitly states that Microsoft reports only 30% AI-generated code, and Source 4 (ShiftMag) shows developers merely "estimate" 42% is AI-assisted, not that companies actually "report" vast majorities as the claim requires. You're essentially arguing that trending toward a minority percentage somehow proves a current vast majority, which is logically incoherent when your best evidence caps out at 42% assisted code.

O
Opponent Argues FALSE

The claim is demonstrably false as Source 4 (ShiftMag) and Source 8 (SonarSource) both report that only 42% of code is AI-assisted, while Source 10 (Fortune) reveals that even at AI-focused companies like Microsoft and Salesforce, only 20-30% of code is AI-generated, falling far short of a "vast majority." Source 12 (METR) further undermines the claim by showing AI tools actually slow developers down by 19%, contradicting the narrative that companies would rely on AI for the majority of their code production.

P
Proponent Rebuttal

You're treating “vast majority” as if it must mean “AI-generated only,” but the motion's wording (“written by AI”) reasonably includes AI-assisted authorship—and the very sources you cite show a near-majority already: developers estimate 42% of committed code is AI-generated or assisted today (Source 4, ShiftMag; Source 8, SonarSource), with expectations of rapid growth toward ~65% by 2027 (Source 4, ShiftMag). And your METR point is a non sequitur: Source 12 (METR) finding a 19% slowdown in a specific study doesn't rebut whether major companies report high AI involvement in code, nor does it negate leadership-reported substantial internal AI code generation like Microsoft's ~30% figure (Source 10, Fortune).

You'll need an account first

Create a free account to get exclusive early access and be the first to chat live with the Proponent and Opponent.

Live Chat is in closed beta

We're rolling this out to a small group first. Join the waitlist and we'll let you know as soon as your access is ready.

The Adjudication

How each panelist evaluated the evidence and arguments

Panelist 1 — The Source Auditor
Focus: Source Reliability & Independence
False
2/10

The most reliable, claim-relevant evidence in the pool is Source 10 (Fortune, citing public executive statements) reporting Microsoft at ~30% AI-generated code and similar figures for Salesforce, while the better survey-style sources (4 ShiftMag; 8 SonarSource; 5 JetBrains; 6 Stack Overflow) indicate widespread AI tool use but place AI-generated/assisted code around ~42%—not a “vast majority”—and none show “major software companies” reporting that most of their code is AI-written. Based on these higher-quality sources, the claim that major software companies report the vast majority of their source code is now written by AI is not supported and is contradicted by the best-anchored company-quote evidence, so the claim is false.

Weakest Sources

Source 18 (DEV Community) is unreliable because it is an opinion/community post making an unsourced prediction (“ninety percent”) rather than reporting verified company disclosures or a methodologically transparent study.Source 14 (Netcorp) is unreliable because it is a marketing/blog-style secondary compilation with unclear methodology and likely circular citations, not primary company reporting.Source 15 (EliteBrains) is unreliable because it appears to be a low-transparency SEO/marketing compilation that asserts global percentages without primary documentation.Source 9 (Anthropic) is potentially conflicted because it is a vendor-produced report with promotional incentives, and the snippet provided does not clearly substantiate the specific claim about “vast majority of source code” across major software companies.
Confidence: 7/10
Panelist 2 — The Logic Examiner
Focus: Inferential Soundness & Fallacies
False
3/10

The evidence shows AI-assisted code ranges from 20-42% at major companies (Sources 4, 8, 10: ShiftMag, SonarSource, Fortune report Microsoft at ~30%, developers estimate 42% AI-assisted), which logically refutes "vast majority" (typically meaning >50-60% and often implying much more). The proponent commits a scope fallacy by conflating "trending upward" with "currently at vast majority" and equivocates between "AI-assisted" and "written by AI"—the claim asserts companies *report* vast majorities *now*, but the highest documented figure is 42% (a minority), and even the proponent's rebuttal concedes this is an "estimate" trending toward 65% by 2027, not a current reality; the claim is therefore false.

Logical Fallacies

Equivocation: Proponent conflates 'AI-assisted' (human writes with AI suggestions) with 'written by AI' (AI generates code), treating them as equivalent when they represent different levels of AI authorship.Scope mismatch / Hasty generalization: Proponent uses current 42% figure and upward trend to conclude 'vast majority' is already achieved, when 42% is definitionally a minority and future projections (65% by 2027) do not prove current state.Moving the goalposts: Proponent redefines 'vast majority' post-hoc to include AI-assisted code and reinterprets 'now' to mean 'in practice' or 'trending toward,' when the claim explicitly states companies report vast majorities currently.
Confidence: 9/10
Panelist 3 — The Context Analyst
Focus: Completeness & Framing
False
3/10

The claim asserts "vast majority" but the most recent evidence shows only 42% of code is AI-assisted (Sources 4, 8), Microsoft reports ~30% (Source 10), and even a Science study found only 29% of GitHub Python functions are AI-written (Source 10)—none of which constitute a "vast majority" (typically >60-70%). The claim cherry-picks the aspirational language around AI adoption trends while omitting the actual quantified percentages reported by major software companies, which consistently fall well short of "vast majority," and conflates "AI-assisted" (where humans write with AI suggestions) with "written by AI" to create a misleading impression of AI autonomy that the evidence does not support.

Missing Context

The highest reported figure is 42% AI-assisted code (Sources 4, 8), not a 'vast majority' which typically means >60-70%Microsoft reports only ~30% AI-generated code and Salesforce similar figures (Source 10), contradicting the claim about major software companiesA Science journal study found only 29% of U.S. GitHub Python functions are AI-written, with lower percentages elsewhere (Source 10)The claim conflates 'AI-assisted' (human-written with AI suggestions) with 'written by AI' (AI-generated), creating a misleading impression of AI autonomyThe 42% figure represents developer estimates of AI-assisted code, not official company reports as the claim statesEven at AI-focused companies like Anthropic and OpenAI, the 100% figure cited (Source 10) refers to individual engineers' workflows, not company-wide averagesProjections show growth to 65% by 2027 (Source 4), indicating current levels are well below 'vast majority' thresholds
Confidence: 9/10

Adjudication Summary

All three evaluation axes converged on "False" with high confidence. Source quality analysis found the most reliable evidence (Fortune reporting Microsoft at ~30%, industry surveys at 42%) directly contradicts the claim. Logic analysis identified the core flaw: 42% is definitionally not a "vast majority." Context analysis revealed the claim cherry-picks aspirational trends while ignoring actual reported percentages that consistently fall short of majority levels.

Consensus

The claim is
False
3/10
Confidence: 8/10 Spread: 1 pts

Sources

Sources used in the analysis

#1 Stanford HAI 2025
NEUTRAL
#2 Deloitte 2026
NEUTRAL
NEUTRAL
#4 ShiftMag 2025
REFUTE
#5 JetBrains Blog 2025-10
NEUTRAL
#6 Stack Overflow 2025
NEUTRAL
#7 DORA 2025
NEUTRAL
#8 SonarSource 2026
SUPPORT
#9 Anthropic 2026
NEUTRAL
#10 Fortune 2026-01-29
REFUTE
#11 Keyhole Software 2026
NEUTRAL
#12 METR 2025-07-10
REFUTE
#13 Zapier 2026
NEUTRAL
#14 Netcorp 2026
SUPPORT
#15 EliteBrains 2026
SUPPORT
#16 Intelegain 2026
REFUTE
NEUTRAL
#18 DEV Community 2026
SUPPORT