TL;DR: Generative AI has made phishing, vishing, and smishing more convincing and more coordinated, pushing Living Security Human Risk Management Platform’s analysis toward multi-vector simulation, identity-aware risk scoring, and continuous reinforcement rather than annual email-only training. The practical shift is from compliance-led awareness to measurable behavior change across the people and access paths attackers actually exploit.
At a glance
What this is: Living Security Human Risk Management Platform says generative AI is pushing social engineering beyond email, making phishing, vishing and smishing too coordinated for awareness programmes built around typos and annual modules.
Why it matters: IAM and security teams need training and measurement that reflect how attackers combine channels, because identity risk is now tied to who is targeted, how they respond and what access they hold.
Context
Phishing awareness has historically focused on a single delivery channel and a single judgement test, usually whether someone clicks a link in email. That model breaks down when attackers can coordinate email, text and voice to create one believable story across multiple touchpoints. For identity programmes, the gap is not just user education. It is governance over how human behavior, access rights and attack timing intersect.
The article’s core problem is that modern social engineering is multi-vector, personalized and increasingly AI-generated, so the old compliance model does not measure the control that matters. In practice, that means teams need visibility across employee behavior, identity context and threat activity, not just completion rates or one-off simulations.
Key questions
A: Security teams should treat phishing as a broader trust problem, not an email problem. Effective training combines simulated attacks with ongoing awareness activities so employees learn the tactics attackers reuse across channels, including urgency, fear, and rewards. The goal is to build pattern recognition that transfers to new messaging platforms, especially as chat tools and cloud integrations expand the attack surface.
Q: Why do AI-generated phishing campaigns increase risk for privileged users and sensitive workflows?
A: AI phishing increases risk because attackers can scale personalised lures that target the right person at the right moment, including users who can approve payments, reset credentials, or access sensitive systems. When a trusted-looking request reaches a high-privilege account, one successful click can cascade into data theft or fraud. Teams should treat privilege as a key risk multiplier, not just user training quality.
Q: What are the signs that phishing awareness training is not working well enough?
A: Training is failing when employees still click, reply, or escalate suspicious messages without checking basic details first. Other warning signs include repeated responses to urgent language, weak reporting rates, and inconsistent use of verification steps for unexpected links or attachments. If people can describe threats but do not change behavior, the program is informative but not operationally effective.
Q: Should awareness programmes prioritise email phishing or multi-channel simulation?
A: Multi-channel simulation should take priority because attackers no longer rely on email alone. Email remains important, but campaigns increasingly combine text and voice to reinforce the same pretext. Training that ignores those channels leaves a false boundary around the threat, while broader simulation reflects how employees are actually targeted.
Technical breakdown
Why single-channel phishing training fails
Email-only training assumes the attack ends where the message lands. In reality, social engineering now crosses channels, with a phishing email followed by a text and then a phone call to reinforce the same pretext. That matters because the attacker is not testing whether someone can spot a typo. They are testing whether the target will accept a coherent story under pressure. Once voice cloning and generative text are added, the old cue-based model becomes fragile because the message quality is no longer the weak point.
Practical implication: design awareness around cross-channel attack chains, not isolated email drills.
How identity context changes human-risk measurement
Human risk becomes more actionable when behavior is joined to identity and access data. A click rate tells you who interacted with a simulation, but it does not tell you whether that person has privileged access, handles sensitive workflows or is already being targeted in an active campaign. Correlating behavior with identity context turns awareness into prioritisation. It lets teams distinguish nuisance exposure from the user whose account could create material blast radius if compromised.
Practical implication: score training outcomes alongside access scope and role criticality, not as a standalone awareness metric.
What AI-native simulation changes in preparedness
AI-native simulation matters because it can reproduce the layered pressure of modern campaigns: a message in one channel, a reinforcement in another and urgency embedded in the final ask. That is different from generic awareness content because it tests recognition under realistic timing and context, not after the fact. It also supports just-in-time reinforcement, where the corrective training follows the risky action closely enough to shape future behavior. The objective is measurable behavior change, not attendance evidence.
Practical implication: replace annual awareness cycles with continuous, scenario-based testing and immediate reinforcement.
Threat narrative
Attacker objective: The attacker wants to convert human trust into access, credential theft or a fraudulent action that bypasses technical controls.
- Entry begins with a personalised phishing message, text or voice contact that establishes a believable pretext and creates initial trust.
- Credential or action capture follows when the target is pressured to click, disclose information, approve a request or hand over a code under urgency.
- Impact occurs when the attacker uses that trust to steal credentials, move into internal systems or trigger fraudulent activity through the compromised human workflow.
Breaches seen in the wild
- MGM Resorts breach 2023: A help desk call gave attackers Okta and Azure admin access at MGM, leading to ransomware, ten days of outages and a $100 million hit.
- Caesars Entertainment breach 2023: Social engineering of an IT support vendor let attackers copy Caesars loyalty database; about $15 million was reportedly paid.
Read and download The State of NHI & AI Agent Breach Report 2026, covering 200+ breaches impacting Non-Human Identities including AI Agents.
NHI Mgmt Group analysis
AI-native awareness is becoming a governance requirement, not a training preference. When attackers can generate convincing email, voice and text in the same campaign, the unit of defence is no longer a single message. The control problem shifts to whether the organisation can detect coordinated persuasion across channels and respond before a user action becomes an incident. Practitioners should treat cross-channel simulation as part of operational resilience, not optional training.
Human-risk programmes are moving toward identity-aware prioritisation. A user’s behaviour only becomes meaningful when paired with access scope, role sensitivity and current threat exposure. That is where the identity bridge matters: social engineering is no longer just a people problem, it is an access problem because a compromised user with privileged reach creates very different risk from a low-impact account. Practitioners should align awareness with IAM and PAM context, not treat all users as equal.
Compliance-led awareness creates false confidence. Completion rates, annual modules and generic reminders tell you that content was assigned, not that risky decisions changed. The article points to a broader market shift toward measurable behavior change, which is the right direction because social engineering now exploits timing, pressure and channel blending rather than simple spelling mistakes. Practitioners should replace checkbox reporting with evidence of reduced risky responses and improved reporting behavior.
Multi-vector social engineering is the new baseline threat pattern. The named concept here is coordinated persuasion, where the attacker uses one channel to validate another and makes each step more believable than the last. That pattern undermines single-vector training and also complicates incident triage, because employees may not realise the messages belong to the same campaign. Practitioners should build controls that assume the attack narrative spans email, SMS and voice.
Behavior-driven security awareness is now the practical benchmark. The value of training is increasingly measured by whether users report, pause and verify under pressure, not by whether they can recite generic advice. That makes continuous reinforcement and targeted scenario design more defensible than broad annual campaigns. Practitioners should expect HRM and IAM teams to share accountability for reducing human-driven access risk.
What this signals
Coordinated persuasion is now the relevant threat model. The practical challenge is no longer whether users can spot obvious fraud, but whether they can resist a campaign that arrives by email, SMS and voice as one narrative. That shifts programme design toward scenario coverage, reporting behavior and escalation paths that work across channels.
Identity context changes how human-risk data should be used. A simulation result only becomes operationally useful when it is tied to access level, role criticality and the likelihood that a bad decision could affect privileged workflows. For practitioners, that means awareness data belongs alongside IAM and PAM signals, not in a separate training dashboard.
The next maturity step is continuous reinforcement, not annual completion. Organisations that want measurable reduction in social engineering exposure need micro-training, realistic simulations and governance that treats human behaviour as part of the access control surface.
For practitioners
- Implement multi-vector simulation programmes Test phishing, vishing and smishing together so users experience the same coordinated pressure attackers use across email, text and voice.
- Correlate awareness results with identity context Combine simulation outcomes with role, access tier and privileged account status so the highest-risk people and workflows are prioritised first.
- Replace annual training with continuous reinforcement Deliver short, just-in-time micro-training immediately after risky actions so the lesson is tied to the actual behavior rather than a later refresher.
- Track reporting behavior, not just clicks Measure whether employees report suspicious messages quickly and consistently, because reporting is a stronger indicator of resilience than completion rates.
Key takeaways
- Generative AI is making phishing, vishing and smishing harder to distinguish from legitimate contact, which exposes the limits of email-only awareness.
- The more useful metric is whether employees report and verify suspicious activity, especially when identity and access context raise the stakes.
- Security teams should treat cross-channel simulation and continuous reinforcement as part of the human-risk control set, not as one-off training content.
Standards & Framework Alignment
This section maps relevant standards and security frameworks to the operational risks and controls described in this guidance.
MITRE ATT&CK addresses the attack and risk surface, while NIST CSF 2.0 and NIST SP 800-53 Rev 5 set the governance and control requirements practitioners need to meet.
| Framework | Control / Reference | Relevance |
|---|---|---|
| MITRE ATT&CK | TA0009;TA0006 — Collection; Credential Access | The article centres on social engineering used to obtain credentials or user actions. |
| Recommendation — Map multi-channel social engineering to TA0006 and TA0009, then monitor for repeated credential or approval capture attempts. | ||
| NIST CSF 2.0 | PR.AT-01 — Security Awareness and Training | This article is fundamentally about modernising awareness and training outcomes. |
| Recommendation — Align awareness metrics to PR.AT-01 and verify training changes behavior, not just completion. | ||
| NIST SP 800-53 Rev 5 | AT-2 — Awareness Training | Awareness training is the primary control family discussed in the article. |
| IR-4 — Incident Handling | Reporting suspicious messages and escalation paths are part of the response model. | |
| AC-6 — Least Privilege | The article stresses that access level changes the impact of a successful lure. | |
| Recommendation — Use AT-2 to structure role-relevant training that covers phishing, vishing and smishing together. Pair awareness programmes with IR-4 playbooks that make employee reporting fast and consistent. Apply AC-6 so a compromised user cannot convert a social engineering win into broad access. | ||
Key terms
- Multi-Channel Social Engineering: Multi-channel social engineering uses more than email to deceive targets, including SMS, voice calls, and impersonation through video or chat. Defending against it requires verification habits and reporting workflows that work across every channel where trust can be manipulated.
- Human Risk Management: The practice of managing how people interact with security controls, especially under pressure, distraction, or deception. It combines training, policy, and friction management so identity systems are still usable enough that users do not bypass them in day-to-day work.
- Just-In-Time Training: Just-in-time training is immediate, short-form guidance delivered at the moment a risky action occurs, such as clicking a simulated phish. It turns a mistake into a learning event while the context is still fresh. In security programmes, this approach improves retention and supports behaviour change better than annual training alone.
- Identity-Aware Risk Scoring: Identity-aware risk scoring is a method for estimating the likelihood and impact of risky behavior by tying signals to a specific identity. It combines authentication context, privilege level, device posture, location, behavior, and resource sensitivity to produce a dynamic score used for access decisions, monitoring, and response.
Deepen your knowledge
NHI Mgmt Group covers identity security, NHI governance, and agentic AI through independent research, practitioner guides, and the NHI Foundation Level course. Explore the NHI Foundation Level course, the industry's only accredited NHI security programme, to connect identity governance to the wider risk patterns your programme depends on.
Published by the NHIMG editorial team on August 21, 2026.
Updated on October 12, 2026.
NHI Mgmt Group, the independent authority on Non-Human Identity, IAM, and Agentic AI security. nhimg.org