Six common identity and access management vulnerabilities and how to fix them
Most cyber breaches don’t start with hackers - they start with weak access controls. Discover six common Identity and Access Management (IAM) mistakes that leave organisations exposed, and learn how strengthening IAM can help reduce risk and improve cyber resilience
by Aurorah Smith Commercial Director

Most organisations still budget for identity as a hygiene task: set a password policy, switch on MFA, run an access review once a year. The attackers have moved on. They increasingly log in rather than break in, and they do it through the parts of identity that hygiene checklists rarely cover: the help desk reset, the admin role nobody removed, the service account nobody owns.
Microsoft's 2026 Digital Defence Report says that most intrusions still start with a person or a credential rather than an exploit and describes identity as 'the primary control plane for defence' (Microsoft, 2026a).
The UK picture is consistent. The Government's Cyber Security Breaches Survey, published in April 2026, found that 65% of medium and 69% of large businesses had experienced a breach or attack in the previous 12 months (Department for Science, Innovation and Technology & Home Office, 2026).
Identity and access management (IAM) is how an organisation makes sure the right people, and the right systems, have the right access to the right resources, and no more.
My view is that identity has moved from hygiene to operating infrastructure, and that 2027 is the year it stops being a technical preference and becomes a dated, budgeted commitment. Passkey migration has deadlines. Certificate lifetimes are shortening. AI is multiplying the number of identities organisations need to govern. And UK regulation is moving further into the technology supply chain. The question for leaders is no longer whether IAM needs attention. It is whether the work is funded, owned and moving quickly enough.
Key takeaways
- Passwords: stolen, reused and breached credentials remain a major route into organisations. Passkeys are now the default authentication experience for Entra users enabled for SMS or voice.
- Phishable MFA and the reset process: SMS codes and push approvals can be bypassed, and so can MFA itself if the help desk will reset it on request.
- Standing privilege: admin rights that are always on turn one compromised account into many.
- Joiners, movers and leavers: accounts outlive the people, contractors or suppliers they were created for.
- Apps and identities outside your control: unsanctioned SaaS, OAuth consents and AI agents nobody owns.
- Weak conditional access and monitoring: every login treated the same, and risky sign-ins unwatched.
Quickest win: check how your help desk verifies a caller before resetting a password or MFA method, starting with admin accounts.
What is identity and access management, and why do attackers target it?
IAM answers two questions. Authentication asks: are you who you say you are? Authorisation asks: now that we know, what are you allowed to do?
The identities in scope are no longer just people. Service accounts, API keys, automation scripts, cloud workloads and, increasingly, AI agents all sign in and act on systems. Microsoft's 2026 report lists the growing use of non-human identities among the things that have changed in the threat picture, while the core techniques of credential reuse, privilege escalation and access misuse are unchanged (Microsoft, 2026a).
A single compromised account is valuable because it rarely stays single. Microsoft found that 52.2% of intrusions using a valid account went on to steal further credentials (Microsoft, 2026a). One identity becomes the route to many, and every step looks like a legitimate user doing legitimate work. That is the commercial point for leaders: identity failures are hard to see, slow to contain and expensive to unwind.
The six vulnerabilities at a glance
1. Weak, reused and breached passwords
What it looks like. Passwords remain the primary way staff sign in. The same password is used across work and personal services, and nobody checks new passwords against known breach lists.
Why it's risky. Stolen credentials remain valuable raw material for identity attacks, and phishing is still one of the most common ways to obtain them. It was the most common attack in the UK Government's 2026 survey, experienced by 38% of businesses (DSIT & Home Office, 2026). It was also identified as the most disruptive type of breach or attack by 69% of businesses and charities that experienced one. Microsoft reports that 89–95% of email phishing attachments led to an effort to steal credentials (Microsoft, 2026a). Reused passwords make every stolen credential worth more: credential stuffing is cheap to run, and it only needs to work once.
How to fix it. Block known breached and common passwords and give staff a password manager so unique passwords are realistic. Then plan the exit: from 1 September 2026, Microsoft made passkeys the default authentication experience for Entra ID users enabled for SMS or voice (Microsoft, 2026b). The direction of travel is passwordless, and the cost of getting there is falling.
Self-check: Do we know what share of our users could sign in tomorrow without a password?
2. Phishable MFA, and the reset process that bypasses it
What it looks like. Multi-Factor Authentication (MFA) is switched on, but it relies on SMS codes, voice calls or push approvals. And when a user calls to say they have a new phone, the help desk resets their MFA after a few security questions.
Why it's risky. MFA that can be phished, or simply reset on request, gives less protection than the dashboard suggests. Many organisations are not yet at the starting line: only 47% of UK businesses use two-factor authentication (DSIT & Home Office, 2026). Attackers are also targeting people rather than technology. Microsoft found that 93% of voice phishing attacks kept the victim on the line long enough to begin social engineering (Microsoft, 2026a), and 12% of UK businesses experienced impersonation in the past year (DSIT & Home Office, 2026). After the attacks on UK retailers in 2025, the NCSC urged organisations to review how their help desks verify staff before resetting passwords, particularly for accounts with elevated privileges (NCSC, 2025a).
There is now a hard date attached. From 1 February 2027, Microsoft stops providing SMS and voice authentication in Entra ID for most users, and users who rely on them alone must register a passkey before they can continue. Global Administrators and external users follow on 1 July 2027 (Microsoft, 2026b). Organisations with a legitimate operational or regulatory requirement to retain SMS or voice will need to use a telephony provider through Microsoft Security Store.
How to fix it. Move privileged users to phishing-resistant MFA, such as passkeys, FIDO2 security keys or Windows Hello for Business, then extend it to everyone. Treat the help desk reset as part of the authentication system: require stronger verification for resets, never reset MFA on a phone call alone, and apply the toughest checks to admin accounts. Microsoft's own guidance notes that authenticator app codes (TOTP) remain supported but are not phishing-resistant (Microsoft, 2026b).
Self-check: If someone rang our help desk claiming to be our IT director with a new phone, what exactly would happen?
3. Excessive privileges and standing admin access
What it looks like. Admin roles are permanent. Access granted for a project is never removed. The number of Global Administrators has crept up because it was quicker than working out the right role.
Why it's risky. Microsoft's 2026 report puts it plainly: 'Attackers consistently take advantage of unnecessary privileges, persistent access, and weak identity governance' (Microsoft, 2026a). In the UK, 73% of businesses say they restrict admin rights (DSIT & Home Office, 2026), which leaves more than a quarter that do not. Standing privilege means a single compromised account can change security settings, create new accounts and disable logging. The NCSC asks organisations to pay specific attention to domain, enterprise and cloud admin accounts and to check that access is legitimate (NCSC, 2025a).
How to fix it. Apply least privilege and make elevated access just-in-time: requested, approved and time-limited. Run regular access reviews that someone is accountable for completing. Keep a small number of tightly controlled break-glass accounts for emergencies, monitored and tested.
Self-check: How many people hold permanent Global Administrator rights today, and could we name each one's reason?
4. Leavers, movers and orphaned accounts
What it looks like. A contractor's project ends but their account stays live. A team member moves to finance and keeps their old access as well as the new. A supplier engineer's account was set up years ago and nobody remembers why.
Why it's risky. Orphaned accounts are back doors that nobody watches, because nobody thinks they exist. Movers are the quieter problem: access accumulates with every role change, so long-serving staff often hold the broadest rights. Supplier and partner accounts carry the same risk with less visibility, and few organisations look closely. Only 15% of UK businesses review the cyber risks posed by their immediate suppliers, rising to 48% of large businesses, and just 6% look at their wider supply chain (DSIT & Home Office, 2026).
How to fix it. Drive joiner, mover and leaver changes from the source of truth, the HR system and contract records, rather than tickets someone has to remember to raise. Give every third-party account an owner and an end date. Audit for dormant accounts on a fixed cycle.
Self-check: When a supplier contract ends, what triggers the removal of that supplier's access?
5. Apps and identities outside your control
What it looks like. Teams sign up for SaaS tools without IT. Users grant third-party apps access to their mailbox or files through OAuth consent prompts. Service accounts and API keys sit in scripts with no named owner. AI agents are being piloted across departments.
Why it's risky. Shadow IT is rarely malicious. People work around systems that feel too slow or restrictive. But every unsanctioned app and unowned identity is access that nobody approved and nobody reviews. Microsoft's 2026 report describes a malicious browser extension, with more than 600,000 installs, that harvested AI chat histories and affected almost 10,000 organisations (Microsoft, 2026a). Gartner found that only 13% of organisations think they have the right AI agent governance in place (Gartner, 2026).
How to fix it. Bring apps under single sign-on so access follows the identity lifecycle. Restrict user consent to third-party apps and route requests through approval. Inventory non-human identities, including AI agents, and give each one a named owner, a defined scope and a retirement date.
Self-check: Could we list every AI agent and service account with access to our data, and who owns each one?
6. Weak conditional access and monitoring
What it looks like. Every login is treated the same, whether it comes from a managed laptop in the office or an unknown device overseas at 2am. Identity risk alerts are generated but nobody acts on them.
Why it's risky. Without context-based rules, a stolen credential works anywhere. Without monitoring, misuse looks like normal work. Microsoft analyses 31 million identity risk detections on an average day (Microsoft, 2026a). The signal exists; the gap is usually in acting on it.
How to fix it. Use conditional access to require stronger authentication, or block access, based on device health, location and sign-in risk. Feed identity events into security monitoring with clear response playbooks. The NCSC specifically recommends watching for risky logins in Microsoft Entra ID Protection (NCSC, 2025a).
Self-check: If an admin account signed in from an unmanaged device tonight, who would know, and how quickly?
Beyond the six: emerging identity risks
The six vulnerabilities above are well understood. What is changing is the scale and type of identity that organisations must govern.
AI agents. Gartner predicts that by 2028 an average global Fortune 500 enterprise will have over 150,000 agents in use, up from fewer than 15 in 2025, and it lists defining agent identity, permissions and lifecycle among its six steps to manage agent sprawl (Gartner, 2026). Microsoft reports that 88% of enterprises are already experimenting with AI agents (Microsoft, 2026a). My read is that most IAM programmes were designed for people who join, move and leave on HR timescales. Agents can be created in minutes, inherit broad permissions and act at machine speed. Governing them is an identity problem before it is an AI problem.
Service accounts, API keys and secrets. These identities can hold more access than any person, and can sit outside the controls applied to people, such as MFA and access reviews. Microsoft found that exposed cloud workloads were attacked within 5.3 hours on average (Microsoft, 2026a).
Third-party access. Suppliers, managed service providers and contractors hold privileged routes into many estates. That is why the UK's regulatory direction, covered below, is moving towards the supply chain.
Token and session theft. Some phishing kits intercept the session after a user has completed MFA, which is why phishing-resistant methods matter. Microsoft reported that activity from one such kit, Tycoon2FA, fell 95% from its November 2025 peak after disruption (Microsoft, 2026a). The technique remains.
Certificates and cryptography. Machine identity also includes the certificates that prove a website or service is genuine. Their lifetimes are shrinking fast, and the move to post-quantum cryptography has begun. Both are covered in the next section.
What changes in 2027
Several developments now put dates around identity and machine-identity planning. The three below are confirmed implementation dates.
|
When |
What changes |
What it means for buyers |
|
1 February 2027 |
Microsoft retires its own SMS and voice authentication in Entra ID for most users; SMS- or voice-only users must register a passkey (Microsoft, 2026b) |
Passkey rollout and user communications need to be finished, or a telephony provider arranged through Microsoft Security Store |
|
15 March 2027 |
Maximum public TLS certificate lifetime falls from 200 to 100 days, and to 47 days from 15 March 2029 (DigiCert, 2025) |
Certificate lifecycle automation becomes increasingly difficult to avoid |
|
1 July 2027 |
SMS and voice retirement extends to Global Administrators and external users (Microsoft, 2026b) |
Global Administrator and external-user access should already be on stronger methods |
Legislation sits on a different footing. As of 2 November 2026, the Cyber Security and Resilience (Network and Information Systems) Bill had reached Third Reading in the House of Lords. Amendments made in the Lords then return to the Commons for consideration before the Bill can progress to Royal Assent (UK Parliament, n.d.; VinciWorks, 2026). As drafted, the Bill would bring many managed service providers within the UK's NIS regulatory regime, with detailed requirements to follow through secondary legislation. Buyers should expect supplier contracts, assurance requests and incident reporting obligations to tighten.
Alongside these, the NCSC asks organisations to complete discovery of their cryptography and an initial post-quantum migration plan by 2028, with the highest-priority migration done by 2031 and full migration by 2035 (NCSC, 2025b). For many larger organisations, that makes 2027 the practical budget cycle in which discovery work needs to be funded if the 2028 milestone is to be credible.
The commercial implication is that IAM work which could previously sit on a 'when we get to it' list now has dates attached. For leaders setting FY27/28 budgets, passkey migration and certificate lifecycle automation are moving from discretionary improvement towards time-bound investment. Supplier access and identity governance deserve to move up the same priority list.
How strong is your IAM? A quick self-assessment
Answer yes or no. Each 'no' is a priority.
- Do all users have MFA, with no exceptions for legacy authentication?
- Are all admin accounts on phishing-resistant MFA, such as passkeys or security keys?
- Does the help desk follow a documented, stronger verification process before resetting passwords or MFA?
- Is admin access granted just-in-time rather than held permanently?
- Are access rights reviewed at least quarterly, with a named owner for each review?
- Do leaver and mover changes flow automatically from HR and contract records?
- Does every service account, API key and AI agent have a named owner?
- Is user consent to third-party apps restricted and reviewed?
- Do conditional access policies consider device, location and sign-in risk?
- Are SMS- and voice-only users on a plan to move before 1 February 2027?
A 30/60/90-day priority view
|
Timeframe |
Focus |
|
First 30 days |
Tighten help desk reset verification; move admins to phishing-resistant MFA; remove unnecessary Global Administrators; restrict app consent |
|
30 to 60 days |
Close MFA gaps and block legacy authentication; deploy risk-based conditional access; start a passkey rollout for SMS- and voice-only users |
|
60 to 90 days |
Automate joiner, mover and leaver changes; introduce just-in-time admin access; inventory non-human identities and certificates with owners |
IAM and compliance
Compliance is not the reason to fix IAM, but regulation increasingly assumes that basic identity controls are already working.
- Cyber Essentials. The NCSC's requirements v3.3 (April 2026) state that 'authentication to cloud services must always use MFA', that cloud services cannot be excluded from scope, and that FIDO2 authenticators such as passkeys count as MFA (NCSC, 2026, Section 4). Only 5% of UK businesses hold Cyber Essentials certification, rising to 35% of large businesses (DSIT & Home Office, 2026).
- ISO/IEC 27001. Its access control requirements cover identity management, authentication, access rights and privileged access.
- UK GDPR. Article 32 requires appropriate security for personal data, and access control is central to showing it.
- Cyber Security and Resilience Bill. As covered above, the Bill would bring many managed service providers into scope. That raises the bar on how suppliers manage their own privileged access into customer environments.
Why IAM is hard to keep on top of in-house
None of the fixes above is technically exotic. The difficulty is sustaining them. Joiners, movers and leavers never stop. Every new SaaS tool and AI pilot adds identities. Identity tooling is often split across directory, privileged access, governance and monitoring products, with the gaps between them owned by nobody. And identity threats do not keep office hours: a help desk social engineering call or a risky admin sign-in at 2am needs a response at 2am.
That is why many organisations look for a partner. If you do, the useful questions are less about tools and more about evidence:
- How do you verify identity before resetting a password or MFA method, for our users and for your own engineers?
- How do your staff access our environment? Ask about just-in-time privileged access, phishing-resistant MFA and session recording.
- How will you get us ready for the February and July 2027 Entra changes, and the move to 100-day certificates?
- How do you discover and govern non-human identities and AI agents, and who owns them?
- What evidence will we see each month that controls are working, not just that they are configured?
- How are you preparing for the Cyber Security and Resilience Bill, and what will change in your incident reporting to us?
How OneAdvanced can help
Where organisations tend to struggle is not knowing what good identity security looks like. It is getting there without disrupting the business, and then keeping the controls effective as people, suppliers, applications and machine identities change.
OneAdvanced IT Services helps customers close that gap across Microsoft Entra, conditional access, privileged access, identity governance and ongoing security monitoring. The aim is an identity control model that can be evidenced, operated and improved over time, rather than another security tool to configure.
Related services: Managed Cybersecurity Services, Modern Workplace Services, Managed Infrastructure, Firewall Management.
OneAdvanced has over 30 years of partnership with Microsoft, so identity improvements are designed around the Microsoft platforms most organisations already run.
Frequently asked questions
What is identity and access management?
The policies, processes and technology that give the right people and systems the right access to the right resources, and no more.
Is MFA enough on its own?
No. SMS, voice and push methods can be phished, and any MFA can be bypassed through a weak reset process. Use phishing-resistant methods alongside conditional access.
What is the principle of least privilege?
Giving each user, service or agent only the access it needs, for only as long as it needs it.
What is conditional access?
Rules that allow, challenge or block a sign-in based on context such as device, location and risk.
How often should access reviews happen?
I would review privileged and sensitive access at least quarterly, and whenever someone's role changes. Higher-risk environments may need more frequent review.
What is an orphaned account, and why is it risky?
An account still active after its owner has left. Nobody uses it legitimately, so misuse can go unnoticed.
Book an identity security health check
Find the identity gaps an attacker would look for first. Start by understanding where your highest-risk gaps sit across authentication, privileged access, joiners, movers and leavers, application consent, machine identities and conditional access, then prioritise what to fix first.
Call us on 0330 343 4000 or request a callback.
About the author
Aurorah Smith
Commercial Director
As Director of Commercial Operations, Aurorah Smith drives strategic growth and operational excellence across the business, ensuring impactful results for clients and stakeholders.
Contact our sales and support teams. We're here to help.
Speak to our expert consultants for personalised advice and recommendations or to book a demo.
Call us on
0330 343 4000Please enter your details, and our team will contact you shortly.
All fields are required
From simple case logging through to live chat, find the solution you need, faster.
Support centre