NABL & ISO/IEC 17025:2017 Calibration Reference Portal
Official Lab Services: multitek.in
Solutions for: Quality Heads & Auditors Shop Floor Inspectors Tool Room Technicians Engineering Students

NABL Software and IT System Testing Laboratory Accreditation 2026: AI, Cybersecurity, VAPT and ISO 17025

 Software, AI and Cybersecurity Accreditation

NABL Software and IT System Testing Laboratory Accreditation 2026: AI, Cybersecurity, VAPT and ISO 17025

Updated: 13 September 2026

Summary: In April 2026, NABL updated NABL 137, “Specific Criteria for Accreditation of Software & IT System Testing Laboratories.” The revised criteria, aligned with ISO/IEC 17025:2017, now explicitly apply to AI/ML, large language models, IoT, blockchain, embedded systems, medical software, cybersecurity, VAPT, cloud security, interoperability, secure coding, and software performance. Accredited labs must show not just technical skill, but also robust test environments, proven staff competence, validated methods, configuration-managed tools, secure data handling, and reliable records.

Today, software testing isn’t just about finding bugs. It’s about proving that digital systems — from medical devices to AI-powered platforms — can perform safely, securely, and as promised. With this shift, the question is no longer just “does it work?” but “how do you know the testing is trustworthy?”

That’s the problem NABL 137 aims to solve for Software and IT System Testing Laboratories.

What’s new in NABL 137 (2026)?

On 29 April 2026, NABL released an updated version of NABL 137. This amendment expands the accreditation criteria to cover today’s complex digital ecosystems.

The current register lists:


  • Document: NABL 137

  • Title: Specific Criteria for Accreditation of Software & IT System Testing Laboratories

  • Issue: 01

  • Original issue date: 14 October 2019

  • Amendment: 02

  • Amendment date: 28 April 2026

The 2026 revision makes these criteria directly relevant to the latest in AI, cybersecurity, cloud, IoT, and more.

What technologies now fall under NABL 137?

The new scope includes, but isn’t limited to:


  • Telecom software and protocols

  • Embedded systems

  • Mobile apps and devices

  • E-governance solutions

  • Data analytics software

  • Medical and clinical software

  • AI/ML systems and large language models

  • Blockchain and cryptography

  • Cyber-physical systems

  • Process automation and control

  • Web applications and cloud platforms

This means NABL 137 is now relevant for far more than just traditional software QA.

Which test parameters are included?

Labs can seek accreditation for parameters such as:


  • Functionality and performance

  • Usability and accessibility

  • Application, network, endpoint, and cloud security

  • Reliability, interoperability, maintainability, portability

  • Code review and conformance testing

Labs should only apply for those areas where they can prove real competence.

Does ISO/IEC 17025 apply to software testing?

Absolutely. NABL 137’s supplementary criteria are built on ISO/IEC 17025:2017 and must be used together.

Although ISO/IEC 17025 is best known for physical testing and calibration, its competence model fits labs performing controlled software and IT-system testing.

Accredited labs must demonstrate:


  • Qualified personnel

  • Sound methods

  • Controlled tools and environments

  • Validated/verified methods

  • Secure, accurate data and reporting

Note: In software testing, “equipment” can mean test tools, cloud platforms, plug-ins, automation suites, and supporting hardware.

Why test tools are treated as equipment

NABL 137 treats software test tools and their environments as equipment, just like physical lab instruments. Changing the tool version, OS, browser, or cloud setup can all affect results — so configuration management is essential.

Configuration management: a lab-quality issue

Labs must be able to show which configuration produced which results. Useful records might include tool names, versions, patches, licenses, dependencies, OS and browser details, user roles, network setup, test data, and suite versions.

Without clear records, repeating a test may not produce the same outcome.

Open-source and cloud tools: allowed, but controlled

NABL 137 allows using open-source or cloud-based tools, provided the lab validates them and records the exact configuration and version at the time of use. The real requirement: understand and control your tools.

Cloud testing: new evidence requirements

With cloud and remote environments, changes can happen without physical access. Labs should document provider details, cloud configuration, region, VM/container setup, network, bandwidth, access controls, data handling, and software versions to ensure results are reproducible.

Test environment separation

Testing must happen in environments separate from production or development. This prevents production activity from skewing results, developers making changes mid-test, test data contaminating live systems, or losing evidence about what was actually tested.

Performance testing: why simulation must be controlled

Performance and load testing often use virtual users, simulated bandwidth, latency, database states, and resource limits. NABL 137 requires labs to document these configurations and explain why the simulation is valid for the test.

VAPT under NABL accreditation

Vulnerability Assessment and Penetration Testing (VAPT) is now explicitly included. Accredited VAPT must use defined methods and scopes, with clear records of targets, techniques, tools, evidence handling, and protection of sensitive data.

Cybersecurity testing: access control is critical

Labs may handle highly sensitive assets, from source code and credentials to attack logs and cryptographic keys. NABL 137 emphasizes strong access controls, secure data handling, and both physical and digital security as part of lab integrity.

AI and machine learning system testing

The 2026 amendment brings AI/ML and large language models into scope. Unlike traditional software, AI systems can behave differently depending on training data, model version, prompt, randomness, and context. Testing requires rigorous control of these variables.

Essential AI test records

Labs should record model name/version, deployment details, test datasets, input prompts, evaluation metrics, acceptance criteria, software dependencies, hardware, randomness settings, observed outputs, and review/approval notes.

Without these, repeating an AI test may yield inconsistent results.

LLM testing: new assessment challenges

Large language models generate outputs dynamically. Labs must move beyond blanket “AI tested” claims, specifying what was tested, how, and against which criteria.

IoT and cyber-physical systems: context is key

Testing must consider firmware, hardware revisions, protocols, sensors, cloud endpoints, mobile apps, security settings, and power states. Focusing only on one piece — like a mobile app — ignores the system’s full complexity.

Medical software: special caution

NABL 137 now includes Software as a Medical Device. Labs must distinguish between accreditation for testing competence and product regulatory approval, medical-device certification, or customer-specific acceptance. Accreditation is not a replacement for regulatory requirements.

Secure coding and code review

Code review is now a formal part of the criteria, covering input handling, authentication, authorization, data exposure, cryptography, dependency risk, error handling, and security-related configurations. Labs must define their review methods and tools.

Accessibility and usability: required quality parameters

Accessibility testing ensures software can be used by people with various abilities. Labs must identify the relevant accessibility standard, tools, manual steps, assistive technologies, and acceptance criteria.

Interoperability testing

Can the system work seamlessly across platforms or organizational boundaries? Interoperability is especially relevant for health, government, finance, industrial IoT, telecom, and cloud environments. Labs must specify protocols, interfaces, data formats, and versions under test.

What does real competence look like?

Real competence isn’t just having experienced developers. It means linking expertise to the actual test area. For example, performance testing needs load modeling and infrastructure skills; VAPT requires security know-how; AI/ML testing demands model and metric expertise; accessibility needs manual review and assistive tech knowledge; embedded systems call for hardware-software integration experience.

Method verification and validation

NABL 137 highlights the need for method verification and validation. Labs must know what requirement each method evaluates, required inputs, tools, outputs, pass/fail criteria, and limitations. Using a popular tool doesn’t automatically mean the method is validated.

Steps to prepare for NABL accreditation

  1. Define your scope. Only include test parameters where your lab has real competence.

  2. Study NABL 137. Ensure every supplementary requirement is mapped and addressed.

  3. Map ISO/IEC 17025. Build your foundational competence framework.

  4. Control your tools. Set clear versions, configurations, licenses, and validation records.

  5. Control test environments. Log infrastructure, access, networking, and simulation.

  6. Demonstrate personnel competence. Link staff to authorized testing activities.

  7. Validate your methods. Ensure repeatable processes and clear acceptance criteria.

  8. Secure your test artefacts. Protect client software and evidence.

  9. Run internal audits. Test your processes with real projects end-to-end.

  10. Use the latest application form. As of August 2026, use NABL 151 Amendment 07.

Common accreditation mistakes

  • Calling a software firm a laboratory without documented, controlled testing

  • Using dynamic cloud environments, but not recording configurations

  • Relying on open-source tools without validation/version control

  • Testing in live production without strict controls

  • Letting developers alter systems during accredited testing

  • Reporting VAPT results without an authorized, defined scope

  • Claiming “AI compliance” without test metrics or acceptance criteria

  • Failing to preserve test data/evidence

  • Issuing unreviewed test-tool outputs

Why this matters in India, 2026

The April 2026 update specifically targets India’s fastest-growing digital sectors: AI, large language models, cloud, cybersecurity, IoT, blockchain, medical software, and embedded systems. This gives technical labs, cybersecurity firms, and independent verification providers a clear, recognized framework for demonstrating their testing competence.

But it also creates a challenge for customers: how do you tell a regular software-testing provider from a lab genuinely accredited for your needs?

The answer: always verify the current NABL certificate and scope — not just a claim on a website.

What customers should look for

  • Is the organization currently NABL accredited?

  • Is your specific software or IT testing need in their accredited scope?

  • Does the scope include the parameter you care about (e.g., security, performance, interoperability)?

  • Does the lab have the right test environment?

  • Does their report specify the software version tested?

  • Are key tools and methods identified?

  • Does the report clearly separate accredited testing from consulting?

The bigger picture: digital conformity assessment

NABL 137 shows that lab accreditation is no longer just for physical testing. Now, digital systems — from AI to cloud to embedded — require reliable evidence for security, performance, interoperability, reliability, accessibility, AI behavior, and hardware-software integration. This is where software engineering, cybersecurity, and metrology overlap, and it’s one of the most important new accreditation opportunities for technical labs in 2026.

FAQ

What is NABL 137?


NABL 137 sets the accreditation criteria for Software & IT System Testing Laboratories. The current edition is Amendment 02, April 2026.


Can a software testing lab be accredited to ISO/IEC 17025?


Yes. NABL 137 works in tandem with ISO/IEC 17025:2017 for these labs.


Does NABL 137 cover AI and large language models?


Yes. The 2026 update explicitly includes AI, ML, and large language models.


Does NABL accreditation cover VAPT labs?


Yes, if VAPT is inside the lab’s accredited scope.


Can accredited testing use cloud or open-source tools?


Yes, provided the lab validates these tools and maintains strict configuration and version records.


Does NABL accreditation mean the software product is approved?


No. Accreditation demonstrates lab competence within a defined scope. Regulatory approval or certification may still require further steps.


Sources