AI in Software Testing Statistics (2026)

Most testing organizations now use AI in some form. Far fewer have it running at scale, and the gap between how fast AI writes code and how fast teams can check that code is now visible in the survey data. Below are the latest numbers from independent surveys, vendor research and SAP user groups, picked for the people who have to sign off on a release.

Key takeaways

Adoption: how widely AI is used in testing

Every major survey finds that most testing teams use AI in some way. The headline numbers range from about 60% to almost 90%, because each survey asks a different question. Some count any use, some count production use, and some count use "across most workflows." Look at the definition before you put one of these numbers in a steering-committee deck.

• In the World Quality Report 2025-26, which surveyed 2,000 executives (Sogeti, 2025), 89% of organizations are piloting or deploying Gen AI–augmented quality engineering workflows: 37% in production and 52% in pilots (Capgemini, November 2025).
• PractiTest's 2026 State of Testing report (vendor) puts AI adoption in testing at 76.8%. 78.8% of respondents name AI as the most impactful trend for the next five years (PractiTest, 2026).
• In Katalon's 2025 State of Software Quality report (vendor), 61% of QA teams are adopting AI-driven testing to automate repetitive tasks (FutureCIO, 2025).
• In Applause's 2025 functional testing survey (vendor), 60% of respondents say their organization uses AI in testing (Applause, 2025).
• BrowserStack (vendor) surveyed more than 250 CTOs, VPs of Engineering and QA leaders in the US, UK and Europe. 61% of their organizations use AI across most of their testing workflows (BrowserStack, 2026).

Source: Capgemini / OpenText / Sogeti, World Quality Report 2025-26, November 2025

The share of organizations not using Gen AI went up between the last two editions of the report. That points to a more careful approach after the first wave of enthusiasm, not to teams abandoning AI.

Source: Capgemini, share of organizations not adopting Gen AI in QE, November 2025

The year before, the World Quality Report 2024-25 (more than 1,750 senior executives in 33 countries) found that 68% of organizations were either using Gen AI (34%) or had roadmaps after successful pilots (34%) (Capgemini, October 2024; sample from OpenText, October 2024).

Analyst forecast (via vendor reprints): Gartner's 2024 Market Guide for AI-Augmented Software-Testing Tools predicts that 80% of enterprises will have integrated AI-augmented testing tools into their software engineering toolchain by 2027, up from about 15% in early 2023 (Katalon, April 2024). The 2022 edition gave the same 80% target but measured it from a 10% baseline in 2022 (Appvance, 2022, historical). The baselines differ, so cite the edition you are quoting.

What teams actually use AI for

Most AI in testing goes into producing and maintaining test assets. Much less goes into deciding what to test. In PractiTest's words, teams use AI as "extra hands" rather than "extra brains."

Source: PractiTest State of Testing (vendor), 2026

• Company size changes the pattern. 65.5% of large firms use AI for script maintenance, compared with 41.7% of startups. Small teams use it more for test planning (50.0% vs 31.0%) and risk identification (33.3% vs 12.1%) (PractiTest, 2026).
• Applause respondents use AI to develop test cases (70%), automate test scripts (55%) and analyze test outcomes (48%). Only 33% say they have comprehensive documentation for test cases and plans (Applause, 2025).
• Katalon reports that 72% of QA professionals use AI for test case and script generation (Katalon, 2025). The same survey's press coverage gives 61% for "adopting AI-driven testing," and Katalon itself puts the sample at "over 1,500" in one place and "over 1,400" in another. Treat these as two different questions, and quote the sample size carefully.
• The World Quality Report finds Gen AI use moving from analyzing outputs, such as defect analysis and reporting, toward shaping inputs. Test case design and requirements refinement now lead adoption (Capgemini, November 2025).

Synthetic test data grew fastest. It is ranked the top Gen AI use case, and its use in testing went from 14% to 25% in one year.

Source: Capgemini World Quality Report 2025-26, 2025

The scaling gap: why pilots stall

The World Quality Report's main finding is the distance between trying AI and running it at scale. 43% of organizations are experimenting with Gen AI in QA, but only 15% have scaled it enterprise-wide (Capgemini, 2025). The reasons have changed. In 2024 the top obstacles were strategic: no validation strategy (50%) and too few AI skills (42%). In 2025 the top obstacles are operational.

Source: Capgemini, World Quality Report 2025-26 press release, November 2025

• 60% of organizations struggle to get secure, scalable test data, and 58% report trouble adopting AI-powered tools (Capgemini, 2025).
• Tool sprawl is a separate problem. 53% of organizations manage 6 to 10 AI or automation tools (Tricentis, 2026), and 33% say that complexity blocks continuous quality at scale (Tricentis, June 2026).

Trust and verification

Use of AI keeps rising while trust in its output falls. That gap matters most to the person signing the release.

Source: Stack Overflow Developer Survey, July 2025; 2024 comparison via BigDATAwire, July 2025

• 84% of developers in the Stack Overflow survey use or plan to use AI tools, up from 76% the year before (Stack Overflow, 2025).
• Conflicting figures: the same Stack Overflow leaders' page says 29% trust AI output to be accurate, down from 40% in 2024. The main survey page reports 33% who trust it. The questions are framed differently, so do not mix them.
• Experienced developers are the most cautious. 2.6% say they highly trust AI output and 20% highly distrust it (Stack Overflow, 2025).
• In Google Cloud's 2025 DORA research (nearly 5,000 technology professionals), 30% have little or no trust in AI-generated code (IT Revolution, September 2025).

Most teams still keep a person in the loop when checking AI-generated tests:

Source: SmartBear State of Software Quality and Testing (vendor), 1,436 US and UK respondents, 2026

Leaders trust AI more than the people doing the testing. In SmartBear's 2026 survey, 72% of leaders trust AI-generated tests more than human-written ones, compared with 56% of practitioners. About 20% of practitioners trust AI tests less, twice the leader rate of 10% (SmartBear, 2026).

Quality and release risk: code is outpacing testing

AI has sped up code production more than it has sped up checking that code. Vendor surveys from 2026 show the gap clearly. Keep in mind that the vendors behind them sell testing tools.

Source: Tricentis Quality Transformation Report (vendor), 2,501 respondents in six countries, June 2026

• The reasons for untested deployments have changed. In 2025, organizations mostly called them accidents. In 2026, 32% point to leadership pressure to put speed ahead of quality, and 30% say there is now too much AI-generated code to test fully (ERP Today, 2026).
• For 20% of organizations, poor software quality costs more than $1 million a year (ExecutiveBiz, June 2026).
• 46% have shipped AI-generated code that later failed in production. 55% had application quality problems in the past year because testing could not keep up with development (SmartBear, September 2026).
• Only 46% of teams validate the specification before AI generates code from it (SmartBear, September 2026).
• In an earlier SmartBear survey of 273 testing and quality decision-makers (January 2026), 70% said application quality had already degraded as AI sped up development (SmartBear, March 2026).
• DORA's 2025 research finds that AI adoption correlates with higher software delivery throughput and also with higher instability: more change failures, more rework and longer time to resolve issues (IT Revolution, September 2025).
• Vendor telemetry: Faros AI's analysis of more than 10,000 developers found AI assistance linked to 98% more pull requests, 91% longer code review time and 9% higher bug rates, while organizational delivery metrics stayed flat (Faros AI, 2025).

Productivity and ROI

People report large productivity gains. Hard ROI is reported by fewer teams.

• 90% of DORA respondents use AI at work, and more than 80% believe it has made them more productive (Google Cloud, September 2025). The median user spends about two hours a day working with AI (IT Revolution, 2025).
• According to Scrum.org's summary of the same report, 59% say AI has improved code quality and 10% say it got worse (Scrum.org, December 2025).
• In the World Quality Report 2024-25, 72% reported faster automation processes after integrating Gen AI (Capgemini, October 2024).
• 45% of developers say debugging AI-generated code takes a lot of time (Stack Overflow via BigDATAwire, July 2025).
• Only 18% of BrowserStack's respondents see returns above 100% from AI in testing. Organizations that have used it for more than four years are 83% more likely to reach that level (BrowserStack (vendor), 2026).

Source: Where AI delivered the most value in software delivery, Tricentis 2026 Quality Transformation Report (vendor), June 2026

Agentic and autonomous testing

Interest in AI agents that generate, run and judge tests is high. Readiness to govern them is low, and the trust numbers point in different directions.

• Conflicting figures, same survey: Tricentis reports that 83% of organizations trust agentic AI to make release decisions (Tricentis press release, June 2026). Its own analysis also shows confidence in AI agents making release decisions falling from 48% in 2025 to 34% in 2026 (Tricentis blog, 2026). The first number most likely includes partial trust and the second full confidence. We report both.
• 82% of organizations feel at least somewhat prepared to govern AI agents. Only 35% feel fully prepared (Tricentis, 2026).
• 36% call AI agent implementation across the SDLC a top IT priority for 2026, and 68% have put AI into some or all delivery workflows (Tricentis, 2026).
• 83% of SmartBear's 2026 respondents say autonomous testing would help them keep up with the volume of AI-generated code (SmartBear, September 2026).
• Only 31% of developers currently use AI agents, and 38% do not plan to. Among those who use agents at work, 69% report higher productivity (Stack Overflow via BigDATAwire, July 2025).

Executives and practitioners rate their readiness very differently:

Source: Tricentis, 2026 and Tricentis press release, June 2026 (vendor)

The same gap shows up in testing strategy. 93% of C-level leaders feel confident in it, while 30% of QA and DevOps leaders do not (Tricentis, 2026).

People and skills

• Generative AI is now the most important skill for quality engineers (63%), just ahead of core quality engineering skills (60%) (Capgemini, 2025).
• 82% of QA professionals think AI skills will be critical within three to five years (Katalon via FutureCIO, 2025).
• 65.6% of testing professionals are "very concerned" about the future of their profession. Testers who actively use AI are 17% less anxious than those who do not (PractiTest, 2026).
• 82% of organizations have dedicated learning pathways for QE teams, but only 50% track whether they work (Capgemini, October 2024).
• Testing is still measured by volume: 56% of teams are judged on test coverage and 4.5% on NPS (PractiTest, 2026).

Source: Capgemini, 2025; PractiTest (vendor), 2026

SAP customers: AI arrives in the middle of a migration

We found no large independent survey that measures AI use in SAP testing on its own. The closest proxies are the SAP user groups' data on AI adoption in general. It shows SAP shops further behind than the overall testing market, with S/4HANA migration work taking budget and attention first.

Source: ASUG AI research with Microsoft and Intel, 142 ASUG members, February 2026, via CIO.com, May 2026

• In ASUG's 2026 Pulse of the SAP Customer survey, 61% of respondents named budget constraints as their top challenge, up 7 points from the previous year. ASUG links that pressure to S/4HANA projects (CIO.com, May 2026).
• Among DSAG members in Germany, Austria and Switzerland, 43% have implemented AI use cases and 51% have not (DSAG Investitionsreport 2026 via All About Security, February 2026).

Source: DSAG members with productive AI use cases, DSAG Investitionsreport 2026 via All About Security, February 2026

For SAP test leads, this means AI in testing will probably arrive through third-party tools and partners before SAP-native capabilities, and on landscapes that are still on-premises or heavily customized.

What this means for your sign-off

• Treat AI-generated tests as unreviewed until someone has reviewed them. 84% of teams already use human review of AI-generated tests, and only 3% rely on AI checking itself. Make that review step something you can audit: log who checked which generated test case and what they changed. Our AI-Generated Test Review Checklist & Log is built for this.
• Fix test data before buying more AI tools. 60% of organizations struggle with secure, scalable test data, and synthetic data is now the top Gen AI use case. In SAP landscapes, data privacy is also the top barrier to scaling (67%). See Test data for S/4HANA: copies, masking and synthetic data.
• Use AI first on regression maintenance. 65.5% of large firms use AI to keep existing scripts working. That is where the S/4HANA upgrade cadence costs the most effort. Start with S/4HANA regression automation: what to automate first.
• Record untested scope as an explicit go/no-go decision. 60% of organizations ship untested code, and 32% cite leadership pressure as the reason. If scope ships untested, write it down in the go/no-go record with a named risk owner, so it does not happen without anyone deciding.
• Keep AI agents out of the approval chain for now. Only 35% of organizations feel fully prepared to govern AI agents, and confidence in agents making release decisions fell to 34%. Set approval gates before agents touch your SAP transports. AI-DLC for ERP teams covers where those gates belong.
• Bring evidence to close the leader–practitioner gap. 93% of C-level leaders feel confident in their testing strategy, while 30% of QA leaders do not. Take coverage, defect leakage and AI-review data to steering committees, not just how confident the team feels.

Frequently asked questions

What percentage of organizations use AI in software testing in 2026?

It depends on the definition. PractiTest's 2026 report puts AI adoption in testing at 76.8% (PractiTest, 2026). The World Quality Report finds 89% piloting or deploying Gen AI in quality engineering, but only 37% in production (Capgemini, 2025).

What is AI most commonly used for in testing?

Producing test assets. 69.6% of teams that use AI apply it to test case creation and 59.6% to script maintenance, compared with 19.9% for risk identification (PractiTest, 2026).

Do testers trust AI-generated tests?

Practitioners trust them less than leaders do. 56% of practitioners trust AI-generated tests more than human-written tests, compared with 72% of leaders (SmartBear, 2026). Among developers in general, 46% distrust the accuracy of AI tools (Stack Overflow, 2025).

Is AI making software quality worse?

The evidence is mixed. DORA links AI adoption to higher throughput and also to more instability (IT Revolution, 2025). In a 2026 vendor survey, 46% of respondents had shipped AI-generated code that later failed in production (SmartBear, 2026).

How far along are SAP customers with AI?

Early. In ASUG's February 2026 study of 142 members, 41% were piloting AI and only 10% had reached enterprise-scale rollout (CIO.com, 2026). Among DSAG members, 43% have implemented AI use cases (All About Security, 2026).

Will AI-augmented testing tools become standard?

Gartner forecasts that 80% of enterprises will have integrated AI-augmented testing tools into their toolchains by 2027, up from about 15% in early 2023 (Gartner via Katalon, 2024).

Sources

1. World Quality Report 2025-26, Capgemini / OpenText / Sogeti, November 2025
2. World Quality Report 2025: AI adoption surges in Quality Engineering, but enterprise-level scaling remains elusive, Capgemini, November 13, 2025
3. World Quality Report 17th Edition 2025-26 (sector report), Sogeti, December 2025
4. World Quality Report 2024 shows 68% of Organisations Now Utilising Gen AI, Capgemini, October 2024
5. World Quality Report 2024 press release, OpenText via PR Newswire, October 22, 2024
6. Katalon recognized in the 2024 Gartner Market Guide for AI-Augmented Software Testing Tools, Katalon, April 2024
7. 2022 Market Guide for AI-Augmented Software Testing Tools, Gartner via Appvance, 2022 (historical)
8. 2025 Stack Overflow Developer Survey, Stack Overflow, July 2025
9. AI adoption is accelerating, but trust is declining, Stack Overflow, 2025
10. Stack Overflow's 2025 Developer Survey Reveals Trust in AI at an All Time Low, BigDATAwire, July 2025
11. Announcing the 2025 DORA Report: State of AI-Assisted Software Development, Google Cloud, September 2025
12. AI's Mirror Effect: How the 2025 DORA Report Reveals Your Organization's True Capabilities, IT Revolution, September 2025
13. DORA Report 2025 Summary, Scrum.org, December 8, 2025
14. Key Takeaways from the DORA Report 2025, Faros AI (vendor), 2025
15. The 2026 State of Testing Report, PractiTest (vendor), 2026
16. Survey explores AI and the future of QA, FutureCIO, 2025
17. Test Automation Statistics & Trends for 2025, Katalon (vendor), 2025
18. 2025 State of Digital Quality in Functional Testing, Applause (vendor), 2025
19. Inside the State of AI in Software Testing 2026, BrowserStack (vendor), 2026
20. The State of Software Quality and Testing 2026, SmartBear (vendor), 2026
21. 46% Have Shipped Failed AI Code, Yet 69% Are Still Confident in It, SmartBear, September 30, 2026
22. SmartBear Survey: 70% of Software Experts Concerned that Application Quality is Suffering, SmartBear, March 18, 2026
23. Tricentis Report: 60% of Global Organizations are Shipping Untested Code, Tricentis (vendor), June 3, 2026
24. New report: We're adopting AI faster than we trust it, Tricentis, 2026
25. 2026 Quality Transformation Report, Tricentis, 2026
26. Tricentis 2026 Quality Transformation Report (PDF), Tricentis via SAPinsider, June 2026
27. AI Made Code Faster—Tricentis Says Quality Teams Are Falling Behind, ERP Today, 2026
28. Tricentis Report Examines Software Quality as AI Use Expands, ExecutiveBiz, June 18, 2026
29. SAP customers say migration is eating their budgets—and AI is next in line, CIO.com (ASUG research), May 2026
30. DSAG-Investitionsreport 2026: SAP bleibt gefragt, All About Security (DSAG data), February 26, 2026

Last updated: October 15, 2026. Statistics are refreshed as new research is published.