/

  / ISO 27001 Background Checks: Annex A 6.1 Explained

ISO 27001 Background Checks: Annex A 6.1 Explained

Most ISO 27001 auditors will tell you the same thing: background checks are one of the most common controls organisations claim to have, and one of the most common controls that falls apart when you ask for evidence. The control itself, Annex A 6.1 (Screening), is short. The discipline of doing it properly across every hire, contractor, and supplier with sensitive access is anything but.

This guide walks through what Annex A 6.1 requires under the 2022 version of the standard, what changed from 2013, how to design a screening process that holds up under audit, and how to apply the control sensibly in small businesses, startups, and AI-first companies where the headcount is low but the data exposure is high.

Critical Role of Background Checks

What Are ISO 27001 Background Checks?

ISO 27001 background checks are pre-employment and ongoing verification activities used to confirm that anyone with access to an organisation’s information assets is who they claim to be, holds the qualifications they say they hold, and presents no known risk that the role would amplify.

The checks are not punitive. They are a control: a way to reduce the probability that a person granted access to sensitive systems turns out to be someone the organisation would never have hired had it asked the right questions up front.

What Does ISO 27001 Say About Background Checks?

The 2022 version of the standard places the requirement in Annex A 6.1, Screening, under the People Controls theme. The control text is brief and worth reading in full:

“Background verification checks on all candidates to become personnel should be carried out prior to joining the organisation and on an ongoing basis taking into consideration applicable laws, regulations and ethics, and be proportional to the business requirements, the classification of the information to be accessed and the perceived risks.”

Three phrases in that sentence carry most of the weight.

“All candidates to become personnel” is broader than employees alone.

“On an ongoing basis” means screening does not end at hire.

“Proportional” gives organisations room to scale checks to risk — and also takes that room away from anyone who wants to apply identical checks to everyone regardless of role.

Reach SOC 2 Compliance in 6 Weeks or Less

Schedule Your Free SOC 2 Assessment Today

ISO 27001:2022 Annex A 6.1 Screening Explained

The control sits in the People Controls section (A.6.1 through A.6.8), which addresses the human factor of information security across the employee lifecycle. Annex A 6.1 is specifically the pre-access gate. By the time a new hire or contractor touches a production system, the screening required for their role should already be complete and documented.

In practice, this means three things: there must be a written screening policy; the policy must be applied consistently; and evidence of completed checks must exist for every person in scope, not just senior or technical roles.

Who Needs to Comply with Annex A 6.1?

Any organisation pursuing or maintaining ISO 27001 certification needs to evaluate this control. It applies to full-time employees, part-time staff, fixed-term contractors, casual or temporary workers, and selected suppliers whose personnel will be granted access to information systems or facilities. Volunteers and interns are typically in scope as well if they handle anything classified above public.

A 6.1 is not a clause from the main body of the standard (clauses 4 through 10), which means organisations can technically exclude it through the Statement of Applicability if they can justify why it is not relevant. In practice, that justification is almost impossible to make for any organisation with employees or contractors handling customer or production data, and auditors are unlikely to accept it.

Insider Note: Auditors rarely challenge whether you do background checks — they challenge whether you can prove you did them consistently. The most common audit finding under A 6.1 is not a missing policy; it is a screening file with a check listed but no completion date, no result, and no evidence that the result was reviewed before access was granted.

Annex A 6.1 Background Checks

When Do Background Checks Need to Be Conducted?

The default is before access is granted. The standard refers to checks being completed prior to joining the organisation, but for ISO 27001 purposes the practical line is whether the person has access to in-scope systems. If you onboard someone administratively two weeks before they touch any sensitive data, that is fine. If you grant production access on day one, screening must be complete on day one.

Re-screening is also expected on an ongoing basis. This does not mean annual criminal record checks for every employee. Still, it does mean defining the events that trigger a fresh check: role changes that increase access, promotions into privileged positions, or significant changes in the regulatory environment.

 

Why Background Checks Matter for ISO 27001 Compliance

The control exists because access is the precondition for almost every information security failure. According to the Verizon 2024 Data Breach Investigations Report, 68% of breaches involved a non-malicious human element — and that figure excludes deliberate insider misuse, which is tracked separately. The people inside the perimeter are not the only risk, but they are the most consistent one.

What Impact Does Employee Access Have on Data Security?

A breach does not require a sophisticated attacker. It requires someone with credentials to do something they should not have been able to do, whether through error, manipulation, or intent. The narrower the gate at hiring, the smaller the population of people who can later become a problem. Background checks do not eliminate insider risk, but they raise the cost of getting a job in your organisation under false pretences, which is the entire point.

How Background Check Results Help You Manage Risk and Stay Compliant

Screening results feed directly into the risk management process required elsewhere in ISO 27001. A failed identity check or an undisclosed serious conviction relevant to the role is not just an HR matter — it is a risk decision the organisation must document. As part of any thorough ISO 27001 gap analysis, auditors will look for evidence that screening outcomes were considered, that decisions were defensible, and that the process did not quietly wave people through when results were inconvenient.

Reach SOC 2 Compliance in 6 Weeks or Less

Schedule Your Free SOC 2 Assessment Today

What Do ISO 27001 Background Checks Cover?

The standard does not prescribe a fixed list. It expects the organisation to define what is appropriate, document the rationale, and apply the checks consistently for roles at equivalent risk levels.

Key Verifications Included in Annex A 6.1 Screening

A defensible baseline for most roles typically includes verification of identity (a government-issued document such as a passport or driving licence), confirmation of right to work in the relevant jurisdiction, verification of claimed academic and professional qualifications, employment history covering the most recent years, and two independent references (one professional, one personal or character).

The CV should be reviewed against the verified record. Gaps and inconsistencies are not automatically disqualifying, but they should be explained, and the explanation should be on file. For roles with access to financial systems, payment data, or regulated information, a basic criminal record check is standard. In the UK, this is a Basic DBS check; equivalents exist in most jurisdictions.

Enhanced Vetting Under Annex A 6.1

Roles with significant privileged access, board-level visibility, or contact with vulnerable populations warrant deeper checks. Enhanced vetting can include credit checks (where lawful and relevant), sanctions and politically exposed persons (PEP) screening, open-source and social media review, and Standard or Enhanced DBS checks (or equivalent) where the role legally qualifies.

Enhanced vetting is not a default. It must be justified by the role and the information classification, and applied consistently to every person in that tier—not selectively.

Handling Incomplete Verifications

Not every check will produce a clean, complete result. References go unreturned. Old employers no longer exist. Qualifications are awarded by institutions that have since closed. The standard does not require perfection; it requires that gaps are identified, alternatives are attempted, and unresolved gaps are documented with a risk decision before access is granted. A screening file that says “reference not obtained, alternative reference accepted, approved by [name]” is fine. A screening file that simply leaves the field blank is not.

Pro Tip: Tier your roles before you design your checks.

Create three or four tiers based on the sensitivity of information accessed and the level of system privilege, define the mandatory and optional checks for each tier, and map every job description to a tier. This is the artefact auditors want to see — and it also stops well-meaning managers from quietly skipping checks on hires they happen to like.

How to Perform ISO 27001 Background Checks

The mechanics matter less than the consistency. Auditors are looking for a process, evidence the process was followed, and a way to demonstrate that exceptions were rare and approved.

Develop a Formal Screening Policy

The policy is the foundation. It should state the scope (who is covered), the tiers (what level of check applies to what role), the lawful basis for processing the data, the retention period for screening records, and the named owner of the process. It should reference applicable employment law, data protection law, and any sector-specific regulation.

Organisations that have identified gaps in their current documentation can benefit from engaging internal audit services to benchmark what they have against what the standard expects.

Stay Consistent Across All Roles

Two people hired into equivalent roles should go through equivalent checks. Inconsistency is the fastest route to both a discrimination complaint and an audit finding.

If a check is mandatory for the role, it is mandatory for everyone in that role — no exceptions for referrals, contractors brought in at short notice, or founders who think the rule doesn’t apply to them.

Run Regular Risk Assessments

Roles change, regulations change, and the threat landscape changes. Review the screening policy at least annually, and trigger an interim review when the organisation adds new categories of data, enters a new regulated market, or significantly changes its access model.

Prioritise Privacy and Data Protection

Background checks involve personal data, and in many jurisdictions, special category data such as criminal convictions. Under GDPR compliance frameworks — including the UK GDPR and the Data Protection Act 2018 — criminal offence data is governed by Article 10 and Schedule 1, and it requires a specific lawful basis, not just consent.

The Information Commissioner’s Office has been clear that consent is rarely the right basis in an employment context, because the power imbalance undermines the “freely given” requirement.

Candidates must be told in advance what will be checked, why, how long the data will be kept, and what their rights are.

Records should be retained only as long as necessary — typically the duration of employment plus a defined retention period — and unsuccessful candidates’ data should be deleted within six to twelve months unless a specific legal basis justifies longer.

Integrate Screening with Your HR Process

Screening should be a step in the standard hiring workflow, not a parallel exercise managed by IT or security in isolation. The HR system should record the tier, the checks required, the completion status, and the date access was approved. Where a third-party screening provider is used, the HR record should still capture the outcome — not just a reference to the vendor’s portal.

Communicate Background Check Requirements to Candidates

Candidates should know at the application stage that the role requires screening and what the screening involves. This avoids late-stage withdrawal of offers, gives candidates the opportunity to disclose anything that might come up, and helps establish the lawful basis for processing the data. Transparency here is not just good practice — it is a legal requirement in most jurisdictions.

Reach SOC 2 Compliance in 6 Weeks or Less

Schedule Your Free SOC 2 Assessment Today

Applicability to Small Businesses, Tech Startups, and AI Companies

Smaller organisations sometimes assume Annex A 6.1 is overkill at their scale. It is not. The control is proportionate by design, but it is still required — and small organisations often have weaker compensating controls, which makes screening more important rather than less.

ISO 27001 Background Checks for Small Businesses

For a 15-person professional services firm, a baseline tier covering identity, right to work, qualification verification, and two references is usually sufficient for non-privileged roles. Anyone with administrative access to client data, finance systems, or production environments should additionally undergo a Basic DBS check (or local equivalent) and a more detailed reference review. The key is proportionality — scaling checks to actual risk rather than defaulting to the same level for everyone or applying nothing at all.

ISO 27001 Background Checks for Tech Startups

Tech startups frequently grant production access broadly because the team is small and the work is fluid. This makes the privileged-access tier larger in proportion than at a larger firm. A pragmatic baseline is to treat every engineer with production access as a privileged hire and apply enhanced checks accordingly. Founders and senior hires should not be exempt — the control applies to everyone, and an unscreened founder is exactly the kind of finding that lingers in audit reports and becomes a case study in what not to do.

ISO 27001 Background Checks for AI Companies

AI companies face a specific variant of the problem. Training data and model weights can be the most valuable assets in the organisation, and the engineers, researchers, and annotators with access to them often work across model development, evaluation, and deployment in ways that blur the usual access boundaries.

Screening should account for the full range of people who can read training data, modify model code, or export weights — including contracted research partners and annotation vendors. Confidentiality and IP assignment terms in the contract are necessary but not sufficient; the screening control sits upstream of them.

Worth Knowing: Insider threat research

Insider threat research — including data summarised in Wikipedia's overview of insider threats and explored in depth by the Cybersecurity and Infrastructure Security Agency (CISA) — consistently finds that insider incidents are most damaging when the actor had been given elevated access without commensurate vetting. The cost of the check is trivial compared to the cost of the incident.

ISO 27001 Background Check Policy: What to Include

A screening policy does not need to be long. It needs to be specific, evidence-based, and consistently applied.

How to Write an Annex A 6.1 Screening Policy

A workable structure has seven sections.

  1. Purpose and scope state what the policy covers and who it applies to.
  2. Roles and responsibilities name the owner (typically the HR lead, with security sign-off).
  3. Tiers and checks define the screening levels and what each includes.
  4. Process describes the workflow from offer to access approval.
  5. Records and retention specify what is kept, where, and for how long.
  6. Legal basis and candidate rights set out the data protection position.
  7. Exceptions and approvals state who can authorise a deviation and how it must be recorded.

Key Considerations When Documenting Your Background Check Process

The policy should reference but not duplicate other documents: the data protection notice given to candidates, the supplier vetting policy, and the access control policy. It should name a review cadence — typically annual — and a trigger list for interim reviews.

It should be approved at the right level of seniority and stored where the auditor can find it without being asked twice. Being aware of the common pitfalls in ISO 27001 documentation at this stage will save significant remediation effort later.

Ownership of Annex A 6.1

The control is most often owned by the HR manager, with information security or compliance providing oversight. This works because HR holds the candidate relationship, the lawful basis, and the records — while security and compliance hold the tiering logic, the risk decisions, and the access approval.

Both functions should be visible in the policy, and neither should be able to approve access without the other’s input, where the role sits in an elevated tier.

 

What Changed from ISO 27001:2013 to ISO 27001:2022

In the 2013 version, screening sat at Annex A 7.1.1 within a 14-domain structure. The 2022 update consolidated the 114 controls of 2013 into 93 controls under four themes (Organisational, People, Physical, Technological), and screening moved to A 6.1 within the People theme.

The substantive requirement has not changed dramatically. The control text was tightened, the requirement for ongoing screening was made more explicit, and the proportionality language was strengthened.

Organisations certified under 2013 and transitioning to 2022 should expect their existing screening process to map across without major rework — but they should refresh the policy to reference the new control number and confirm that ongoing-screening triggers are explicitly documented rather than implied.

Reach SOC 2 Compliance in 6 Weeks or Less

Schedule Your Free SOC 2 Assessment Today

Related ISO 27001:2022 Controls Relevant to Background Checks

Screening does not stand alone. Several other Annex A controls connect to it directly, and understanding those relationships helps when mapping evidence across your ISMS documentation.

A 6.2 — Terms and Conditions of Employment

Employment contracts must state the information security responsibilities of the role. Screening confirms suitability for the role; the contract documents what the role then requires of the person filling it. Both need to exist, and both need to be in place before access is granted.

A 6.3 — Information Security Awareness, Education, and Training

Screening establishes a baseline of trust at hire. Training maintains and builds on it through the employment relationship. Both are required by the standard, and neither replaces the other — a well-screened employee who receives no security awareness training is still a risk.

A 6.6 — Confidentiality or Non-Disclosure Agreements

NDAs and confidentiality terms apply to everyone in scope of screening, including contractors and selected suppliers. The screening record and the signed NDA together evidence that the organisation has done what it can to protect information at the human layer.

A 5.15 — Access Control

Access control assumes the person being granted access has been vetted appropriately. Annex A 5.15 and A 6.1 are paired controls in practice: one defines who can have access, the other confirms the person is suitable to be that who.

An access control policy that grants privileged access without reference to screening status is incomplete regardless of how sophisticated the technical controls are.

Frequently Asked Questions: ISO 27001 Background Checks

Are Background Checks Mandatory for ISO 27001 Certification?

A 6.1 can technically be excluded through the Statement of Applicability, but in practice, it almost never is. Auditors will expect a documented justification for exclusion, and most organisations with employees handling in-scope data cannot make that justification credibly. Treat the control as mandatory unless you have a very specific and documented reason why it does not apply.

Every role with access to in-scope information or systems. The level of check varies with the risk tier, but the requirement to screen does not skip categories of staff. There is no carve-out for short-term contractors, interns, or senior leadership — if anything, those groups often warrant closer attention.

There is no fixed interval mandated by the standard. The control requires ongoing screening, but this can be event-driven — role changes, promotions to privileged access, changes in regulatory requirements — rather than purely time-driven.

Some organisations re-verify identity and right to work annually; others rely on defined trigger events. Either approach is acceptable provided it is documented and applied consistently.

Yes, for any contractor or third-party personnel granted access to in-scope information or systems. The depth of the check can be delegated to the supplier under contract, but the obligation to confirm that screening was completed sits with the certified organisation. “Our supplier handles that” is not sufficient without evidence that the supplier actually did it.

Maintain a screening register that records, for each person in scope: the role tier, the checks required, the date each check was completed, the outcome, the name of the reviewer, and the date access was approved.

The underlying evidence — signed reference letters, DBS certificates, qualification verifications — should be stored securely with appropriate access controls and retention rules applied.

Adverse findings should never trigger automatic disqualification. The standard, and the Rehabilitation of Offenders Act 1974 in the UK along with comparable laws elsewhere, requires that adverse information is assessed for relevance to the role. A historical conviction unrelated to the responsibilities of the position is rarely a basis for refusal. A pattern of dishonesty in a role with financial access is a different matter. The decision, the reasoning, and the approval should all be documented.

Important: Blanket policies that exclude all candidates with any adverse finding are unlawful in most jurisdictions and are an audit finding waiting to happen. The control requires a defensible, individual assessment — not a blanket filter.

Closing Note

Annex A 6.1 is a short control that asks something straightforward: confirm the people you are about to trust are people worth trusting, document what you checked, and keep doing it as roles and risks change.

The organisations that pass this control cleanly are not the ones with the most exhaustive checks. They are the ones who decided what proportionate means for them, wrote it down, and applied it the same way to every hire.

Axipro Author

Picture of Pedro Dias

Pedro Dias

Pedro has been writing online for over 10 years. With experience in all things programming, cyber security, and compliance, he is our editor-in-chief at Axipro.

Blog Highlights

Explore More Articles

Compliance software collects the evidence. A consultant builds the system that evidence is meant to prove. That’s the real difference in the ISO 27001 consultant vs software decision, and most teams only figure it out after they’ve bought one and realized they still need the other. Below, we compare what each route covers, where it breaks down, and what it costs you in time, money, and your team’s hours. Short version: software on its own works for a small group of companies. For most SaaS and tech scale-ups trying to get an enterprise deal over the line, consultant-led implementation on a compliance platform is the faster and safer path to a certificate. Quick Answer: Consultant, Software, or Both? Software-only works if you already have an in-house security lead who’s taken a company through ISO/IEC 27001 before and has the time to own the project. Consultant-only still makes sense if you run mostly on-premise or legacy systems that platforms barely integrate with. For everyone else, which means most cloud-native companies under a few hundred people, a hybrid works best: a platform to handle evidence and monitoring, and a consultant to build the management system and stand behind it in front of an auditor. Here’s why. What an ISO 27001 Consultant Handles ISO/IEC 27001:2022 is a management system standard. Clauses 4 to 10 cover how you run information security, and Annex A lists 93 controls you pick from based on risk. Almost none of it is box-ticking. Most of it comes down to judgment calls about your business, and that’s what you’re paying a consultant for. Scoping, Gap Analysis and Risk Assessment Scope is the first decision you make, and the most expensive one to get wrong. Go too wide and you’ll spend months on controls for systems no customer asks about. Go too narrow and the certificate won’t get through the procurement review it was supposed to pass. A consultant scopes around the deals you’re trying to close, runs a gap analysis, and builds a risk assessment based on your real assets and threats. That’s the document auditors dig into hardest. ISMS Documentation and Policy Writing The standard asks for a specific set of documents: the ISMS scope, information security policy, risk assessment and treatment methodology, Statement of Applicability, risk treatment plan, and evidence of competence, monitoring, internal audit, and management review. A consultant writes these around how your company works day to day, instead of how a template imagines it works. Auditors check whether you follow your own procedures, so a mismatch shows up fast. Internal Audit and Certification Audit Support You need an internal audit before certification, and Clause 9.2 says the auditor has to be objective and impartial. In a small company, the people who built the ISMS can’t credibly audit it, so most teams outsource it through ISO 27001 internal audit services. A good consultant also gets your team ready for the Stage 1 and Stage 2 audits, joins the conversations that matter, and handles corrective actions if the auditor raises nonconformities.  What ISO 27001 Compliance Software Handles Compliance automation platforms, often called GRC platforms, have changed how cloud-native companies get certified. They’re very good at the repetitive, evidence-heavy side of the work. Automated Evidence Collection and Continuous Control Monitoring The platform plugs into your cloud provider, identity provider, code repos, HR system, and device management tools, then pulls evidence on its own. It’ll flag an unencrypted storage bucket, an ex-employee who still has access, or a laptop without disk encryption. For technical controls, that saves weeks of screenshots and spreadsheet tracking. Policy Templates and Annex A Control Mapping Most platforms come with a policy library and map each control to the ISO 27001 clauses and Annex A. You get a starting point and a clear view of which controls have evidence and which don’t. Auditor Access and Ongoing Compliance Tracking Auditors can log in and review evidence themselves, which cuts down fieldwork. After you’re certified, dashboards show when controls slip between surveillance audits, so you aren’t rebuilding evidence from scratch every year. Where Each Approach Falls Short Neither route covers everything by itself. The good news is that the ways each one fails are predictable, so you can plan around them. Limits of Compliance Automation Platforms A platform can tell you a control is failing. It can’t decide your scope, run your risk assessment, write a policy that matches your operations, convince your CTO to change the offboarding process, or explain to an auditor why you excluded a control from your Statement of Applicability. Templates can also make you feel further along than you are. A dashboard at 90% can hide an ISMS that won’t survive Stage 1, because the missing 10% is the management system itself. Insider Note: The Stage 1 problem we see most on software-only projects is a risk assessment copied straight from the platform’s default risk library. The risks are generic, the scores are almost identical, and nothing ties back to the company’s own assets. Auditors notice within minutes, and it weakens the Statement of Applicability that’s built on it. The other problem is ownership. Software assumes someone inside the company will drive the project. At most startups that’s a CTO or ops lead who already has a full-time job, and the subscription renews whether the work gets done or not. Limits of a Consultant-Only Approach A consultant working without automation spends billable days on things a platform does for free, like chasing screenshots, updating evidence trackers, and collecting the same proof again before every surveillance audit. You pay more and wait longer. You also end up with a program that’s only accurate on the day it’s handed over. Once the engagement ends, the evidence goes stale and year-two surveillance turns into a scramble. ISO 27001 Consultant vs Software: Side-by-Side Comparison Factor Consultant only Software only Hybrid (consultant + platform) Time to audit readiness 3 to 6+ months Highly variable; depends on internal expertise As little as 6 weeks for well-scoped

Uzbekistan regulates artificial intelligence through two documents. The first is Law ZRU-1115, signed on 21 January 2026. It amends existing legislation to define AI, stops anyone from basing decisions about people’s rights on AI output alone, and fines companies that process personal data unlawfully with AI. The second is the set of Ethical Rules approved by Order No. 3787, in force since 17 June 2026, which spell out what developers, implementers, and users actually have to do. Uzbekistan hasn’t passed a standalone AI act, and its rules don’t sort systems into risk tiers or require conformity assessments. The framework is short and blunt, and it’s already enforceable. Below we walk through what each document requires, who it applies to, how it stacks up against the EU AI Act, and what a company using AI in Uzbekistan should do next. Uzbekistan AI Regulation at a Glance (TL;DR) Instrument Date What it does Who it binds Law ZRU-1115 Signed 21 January 2026 Defines AI in law, sets general rules for AI-built information resources and systems, bans legally significant decisions based only on AI, adds fines for unlawful AI processing of personal data State bodies, organizations, website owners, anyone processing personal data with AI Order No. 3787 (Ethical Rules) Registered 14 March 2026, in force 17 June 2026 Sets eight mandatory ethical principles and lists rights and obligations for developers, implementers, and users Individuals and companies developing, implementing, or using AI in Uzbekistan Law No. 1125 (Personal Data amendments) Adopted 26 March 2026 Limits data localization to biometric, genetic, and local telecom user data, and allows cross-border transfers under conditions Personal data operators, including AI providers AI Strategy until 2030 (RP-358) 14 October 2024 Sets national targets for AI adoption, infrastructure, and skills Government bodies What Is Law ZRU-1115? The law’s official title is a mouthful: “On making additions and changes to certain legislative acts of the Republic of Uzbekistan in connection with the regulation of relations arising from the use of artificial intelligence.” Put simply, it’s an amending law. Instead of creating a new AI code, it writes AI into laws that were already on the books. When It Was Signed and When It Took Effect The Legislative Chamber of the Oliy Majlis adopted the bill on 12 August 2025, and the Senate approved it on 1 November 2025. President Shavkat Mirziyoyev signed it on 21 January 2026. You can read the official text in Lex.uz, Uzbekistan’s national legislation database. The law set out the principles and the penalties. The day-to-day detail arrived later with the Ethical Rules, which came into force on 17 June 2026. For compliance planning, treat mid-June 2026 as the point when the whole framework started applying. Why Uzbekistan Amended Existing Laws Instead of Passing a Standalone AI Act Uzbekistan wants more AI, not less. Its national strategy sets numeric targets for adoption, investment, and local computing capacity, and a heavy EU-style act would have worked against them. So lawmakers kept it light. They defined AI, drew two hard lines (human control over decisions that affect people’s rights, and protection of personal data), and left the Ministry of Digital Technologies to fill in the rest through secondary rules. Businesses get less legal certainty, and the government gets to move faster. Which Laws ZRU-1115 Changes For businesses, two amendments matter most. The Law “On Informatization” (ZRU-560-II, 2003) now contains a legal definition of AI, a new article on using AI in information resources and systems, duties for website owners, and updated powers for the ministry in charge. The Code on Administrative Liability now includes an offense for processing and spreading personal data unlawfully using AI. The Legal Definition of Artificial Intelligence in Uzbekistan Under the amended Law “On Informatization,” AI is a set of technological solutions that imitate human cognitive functions, including learning on their own and solving problems, and that produce results on specific tasks comparable to what a person could do. That’s deliberately broad. It covers generative AI, machine learning classifiers, recommendation engines, and most agentic systems. The Ethical Rules add a narrower term, the AI system: software built on AI that can find, collect, store, analyze, process, evaluate, and use data, and make decisions on its own based on that data. If your product makes a decision from data, or shapes one, assume it counts. Key Rules Introduced by Law ZRU-1115 General Principles for Using AI in Information Systems and Resources The new article in the Law “On Informatization” starts from harm. Information resources created with AI, and information systems running on AI, must not harm people’s life, health, freedom, honor, or dignity, or violate their other inalienable rights. The standard is short and open-ended. It gives regulators something to enforce against without saying in advance what counts as harm. Principle-based rules like this deserve to be taken seriously precisely because the edges are undefined. Human Oversight: No Decisions on Rights and Freedoms Based Solely on AI Most coverage leads with this provision, and it’s easy to see why. When someone makes a legally significant decision that affects human rights and freedoms, they can’t rely only on conclusions produced by AI systems or AI-built information resources. AI can feed into the decision, but a person has to make it. That applies to loan denials, benefit eligibility, hiring rejections, licensing outcomes, and disciplinary action. In each case, someone needs to look at the AI output and own the final call. Insider Note: In AI governance engagements, teams rarely struggle to show that a review step exists. What they struggle to show is that the reviewer could disagree, and sometimes did. If a human clicks “approve” on every AI recommendation and nobody ever records an override, auditors will see automation with a signature on top. Build the override path and log when people use it, starting on day one. Powers of the Authorized State Body (Ministry of Digital Technologies) ZRU-1115 makes the Ministry of Digital Technologies the authorized state body for AI. Among its new jobs, it’s

You can get a SaaS company ready for a SOC 2 audit in six weeks, but you’ll feel every one of them. Most published timelines say three to six months. For a company with no project owner, no identity provider, and nothing written down, that’s about right. A cloud-native startup that already has the basics in place and can protect some time is a different story, and it can fit the work into six hard weeks. This plan walks through that route one week at a time. Each week has an owner, an hour estimate, and a clear test for when it’s finished. The free Google Sheet version turns the plan into a tracker you can hand out to owners and update in your weekly standup. Before you start, know what you’re signing up for. At the end of week 6 you’ll be audit-ready, which isn’t the same as holding a Type II report. Nobody can get you a Type II in six weeks. This is also the do-it-yourself route, and it takes a lot of hours. We’ll show you where those hours go and what the faster option looks like. Is Six Weeks Realistic for Your Company? Six weeks works when most of the plumbing already exists and your job is to formalize it, fill the gaps, and prove it all works. It falls apart when you’re building the foundations and documenting them at the same time. Go through this table honestly before you promise a customer a date. Six weeks is realistic if… Plan for 10 to 16 weeks if… Your product runs on a major cloud provider You host on-premise or across several data centers You already use an identity provider with SSO Every tool has its own login and password You have fewer than about 50 employees You have multiple offices, subsidiaries, or products in scope One named person owns the project with 10 to 15 hours a week Compliance is “everyone’s job,” so in practice nobody owns it An engineer can give you 15 to 20 hours in weeks 3 and 4 Engineering is fully committed to a launch You only need the Security criteria You need Availability, Confidentiality, or Privacy on day one Landing mostly in the right-hand column doesn’t mean you should throw the plan out. Give each week two weeks instead of one and follow the same order. What “SOC 2 Ready” Means at the End of Week 6 SOC 2 doesn’t give you a certificate. An independent CPA firm examines your controls against the AICPA Trust Services Criteria and writes a report, and which of the two report types you go for decides what you can show a buyer after week 6. A Type I report checks whether your controls are designed properly on a single date. Once you’re ready, a Type I audit can start almost right away. A Type II report checks whether those controls kept working over an observation period of at least three months, and usually six to twelve. Most enterprise procurement teams want Type II in the end. Being “ready” at the end of this plan means your in-scope controls are in place, you can pull evidence for any of them on request, and your auditor is booked. From there you either start a Type I audit or open your Type II observation window. Plenty of buyers will sign with a Type I report plus a letter from your auditor saying the Type II period is underway. Important: The Type II clock doesn’t start until your controls are running. If readiness slips by a week, your Type II report slips by a week too. Founders who tell a prospect “we’ll have SOC 2 in Q3” often forget this and end up renegotiating the deal. Before Week 1: Four Decisions to Make First Settle these before the clock starts. If you change any of them halfway through, you’ll redo work. Scope. Decide which systems, teams, and data the report covers. For most SaaS companies that’s the production environment, the code repository, the identity provider, customer data stores, and any support tools that touch customer data. Corporate systems that never see customer data can usually stay out. Trust Services Criteria. Security (also called the Common Criteria) is mandatory. Availability, Confidentiality, Processing Integrity, and Privacy are optional. Report type. Pick Type I if a deal is blocked right now and the buyer will accept it. If there’s no deadline, go straight to Type II. You’ll need it eventually, and skipping Type I saves you an audit fee. Owner and tooling. Name one person who’s accountable for the plan, and decide where your controls and evidence will live. The tooling choice gets its own section below. Pro Tip: Adding Criteria Only add optional criteria when a customer contract or security questionnaire asks for them. Each one brings more controls to set up and more evidence to collect, and you can widen the scope in next year’s audit. Spreadsheet or Compliance Software: Choosing Your Tracking Tool Every SOC 2 program needs a system of record, meaning one place where each control, its owner, its status, and its evidence live. You can run it yourself in a spreadsheet or a GRC platform, or have a consultant implement it for you. The right choice depends mostly on which report you’re after and how much of your team’s time you can spare. A spreadsheet is free and familiar. It also makes you understand your own environment before you automate any of it. For a Type I, or for a small team with a tight scope, a well-built spreadsheet can take you all the way to the audit. Axipro’s free GRC workbook for SOC 2 and ISO 27001 covers all 33 SOC 2 Common Criteria plus the optional criteria, with evidence, risk, policy, and gap trackers built in. It has no macros and opens straight in Google Sheets or Excel. A GRC platform connects to your cloud, identity provider, code repository, and HR system.