SMB1001:2027 changes: what is new and what to expect when you certify
November 21, 2026
SMB1001 is updated every year, and the 2027 edition is the biggest change since the standard launched. We have compared SMB1001:2027 with SMB1001:2026 control by control. This article summarises what has changed in plain language and what it means if you are certifying or renewing in 2027.
The short version
The standard grows from 39 to 44 controls. Bronze and Silver stay at 7 and 17. Gold goes from 27 to 28, Platinum from 32 to 37 and Diamond from 39 to 44.
Platinum carries the biggest increase. Application control moves down from Diamond, and Platinum adds mobile device management, network segmentation, an AI governance framework and incident response exercises.
Gold adds one control: an annual tabletop exercise to test your incident response plan.
Diamond adds phishing-resistant MFA for admin accounts and technical controls to stop unapproved AI tools.
Almost every tier gets tighter wording. Unsupported software, cloud and provider-managed servers, AI policies, asset registers and incident response plans all have clearer requirements.
Changes at every tier (Bronze and up)
Patching now covers unsupported software
The patching control now says software, operating systems and hardware must stay within the vendor's support life. Anything that has reached end of life or end of support must be replaced, upgraded or given a formal business exception with documented compensating controls. Old Windows versions, unsupported routers and ageing NAS devices are the usual suspects.
There is also a new route for operational technology. If a connected machine or industrial system cannot be patched without risking production, safety or warranty, you can document it, explain why and record the compensating controls, such as segmentation or extra monitoring. Each exception must be reviewed at least once a year.
Passwords: passphrases and breach checks
The password control now expects a policy based on long, unique passphrases that have not appeared in known data breaches. If you do use password expiry, passwords must change at least every 12 months.
Anti-malware covers all endpoints
The control is now called "anti-malware" rather than "antivirus" and covers all endpoints as defined in the standard's glossary. In practice the requirement is the same: actively updated protection on computers, and official app stores with built-in protection switched on for phones and tablets.
Silver (Level 2): clearer rules for real-world setups
Provider-managed servers. If a hosting or cloud provider patches your servers under a shared-responsibility model, you can now show compliance with written confirmation from the provider, a current security certification or their published trust-centre documentation. You also need a simple record of which servers are yours and which are managed by a provider.
Individual accounts, with a sensible exception. Shared logins are still not allowed. But kiosks, single-purpose shared terminals and emergency break-glass admin accounts can now use credentials held in a password manager, provided access is approved or time-limited, logged per user and protected by MFA.
Visitor register. This now applies where physical security is a meaningful risk to your business, and the register no longer needs a signature.
Gold (Level 3): one new control and stronger plans
New: test your incident response plan every year
In 2026, incident response exercises only appeared at Diamond. In 2027, Gold requires an annual tabletop exercise. The team walks through one realistic incident, confirms who does what and who has authority, and identifies at least one improvement to the plan. You must record the date, scenario, participants and improvements.
Tighter requirements on existing Gold controls
Cyber security policy: must name the executive, director, owner or principal who is accountable for cyber security.
Incident response plan: must name who has authority to direct the response, and should list the logs you would need to investigate (sign-in, endpoint, email and cloud logs) and how long you keep them.
AI policy: now more practical. Staff who use AI tools must acknowledge the policy in writing. It should cover acceptable uses, what data must not go into AI tools and how to handle AI output that influences business decisions. You also need a list of approved AI tools shared with staff, and an annual review.
Asset register: changes from a register of where important data is stored to a register of IT and physical assets, such as servers, laptops, phones and network equipment. Businesses that run physical processes should also list control systems, such as building management or industrial controllers. It must be updated as assets change.
DMARC: still requires p=quarantine or p=reject, but alignment is now required with SPF or DKIM rather than both, which matches how DMARC actually works.
Cyber insurance: still required. The note allowing an exception where cyber insurance is not commercially available no longer appears in the control text, so check with your certifier if this affects you.
Password managers: the standard now spells out that sharing is for service accounts, API keys and break-glass accounts, not for routinely sharing staff logins.
Platinum (Level 4): five new controls
Platinum goes from 32 to 37 controls, so this is where most businesses will need the most preparation.
Application control moves down from Diamond. Workstations and laptops, including personal devices used for work, may only run an approved list of applications, enforced technically. The list must be documented and reviewed every year.
Mobile device management (MDM) is new. You need MDM or equivalent endpoint management to enforce and report on encryption, device compliance as a condition of access, minimum operating system versions, screen locks, configuration baselines and remote lock or wipe. For personal devices, it should apply only to the work data or work profile.
Network segmentation is new. Keep IT, IoT and OT on separate networks, keep guest Wi-Fi to internet-only access, and block network ports that are not in use. Smart TVs, cameras, printers and building systems are the common gaps.
An AI governance framework is new and goes beyond the AI policy. It covers approved and prohibited uses, human oversight of AI-driven decisions, guardrails for AI agents that can act on their own, a named accountable person, data governance, AI risks and monitoring, reviewed at least annually.
Incident response exercises are more demanding. You need at least two scenarios a year, one of them ransomware or a data breach, and the person with decision authority must take part in at least one.
One existing Platinum control is also clarified: MFA on remote desktop (RDP) now explicitly covers RDP access provided by an outside vendor, where you need written confirmation or a security attestation that the vendor enforces MFA.
Diamond (Level 5): the highest bar
Diamond includes all the Platinum changes, plus:
Phishing-resistant MFA for every admin and privileged account, using FIDO2 or WebAuthn security keys, device-bound certificates or passkeys. Codes, SMS, voice calls and push-only approvals are not enough on their own. Break-glass and service accounts that cannot support it need documented compensating controls and closer monitoring.
Technical controls on unapproved AI. You must block AI tools that are not on your approved list, or stop them being used with company data, log attempts and review those logs at least quarterly. Exceptions need a justification, an approver and an expiry date.
Stricter application control: the requirement becomes mandatory wording ("shall") at Diamond.
Technical incident exercises: on top of the tabletop scenarios, at least one functional or technical exercise each year, such as a red, blue or purple team exercise, ideally run by an outside specialist.
Penetration testing must be human-led. Automated scanning alone no longer counts. The scope must be agreed in writing beforehand, and findings must have a remediation plan tracked to closure.
Police checks extend to directors and founders who hold privileged access, not just employees and contractors.
What to expect when you certify to SMB1001:2027
Bronze and Silver: no new controls, but check for end-of-life software and devices, and update your password policy to passphrases. If you use cloud or hosted servers, collect your provider's security attestation now.
Gold: plan a one-hour tabletop exercise and record it. Add the accountable executive to your cyber security policy, update your incident response plan with decision authority and log sources, refresh your AI policy with an approved tools list and written acknowledgements, and rebuild your asset register to cover devices and equipment.
Platinum: start early. Application control, MDM and network segmentation all take time to roll out without disrupting staff. Budget for an MDM platform if you do not have one, and plan an AI governance framework.
Diamond: plan for security keys or passkeys for every admin, a way to block unapproved AI tools, a technical incident exercise and a human-led penetration test.
Timing: your current certificate stays valid until it expires. Check with your certifier which edition your renewal will use, and run a gap assessment against 2027 well before your renewal date.
How CyberGrape helps
CyberGrape is a CyberCert Partner and SMB1001 Gold certified, and the CyberGrape GRC platform supports both SMB1001:2026 and SMB1001:2027. We can map your current controls and evidence to the 2027 edition, show exactly which new requirements apply at your tier, and help you close them before renewal. Take the free readiness assessment or talk to us about moving to SMB1001:2027.