How do malicious GenAI services change phishing risk?

How malicious GenAI services change phishing tradecraft

Malicious GenAI services reduce the cost and time needed to produce believable phishing at scale. That changes phishing from a labour-intensive craft into a faster, more industrialised workflow, which increases volume, local relevance and message variety. The core risk is not just better wording, but more lures that can be tailored to a target’s role, language, timing and context.

Because phishing still depends on social trust, the practical effect is often a higher baseline of “good enough” lures rather than a sudden leap to perfect deception. That matters because defenders should expect more campaigns that look routine at first glance, then slip past weak review processes, template-based filters or training that only recognises generic scam language.

For a broader view of phishing techniques, lure patterns and layered defences, see Phishing Guide.

Why scale and personalisation matter more than raw sophistication

GenAI changes the attacker’s economics. A phishing operator can generate many variants, test them quickly and adjust the tone, language and pretext for different audiences without spending the same manual effort on each message. That raises campaign throughput and makes low-effort attacks more credible, especially when they exploit current events, internal jargon or job-specific workflows.

Personalisation also increases the chance that the message will feel locally plausible. A target may not be fooled because the phishing is technically advanced, but because it is contextually familiar enough to invite a hurried click, reply or credential entry. In practice, that means phishing success can rise even when the underlying delivery methods stay conventional.

Malicious GenAI services can also improve consistency across large campaigns, which lets attackers iterate faster on subject lines, body copy and call-to-action phrasing. That is why the defensive problem is not limited to spotting a single suspicious email, but to dealing with repeated, adaptive attempts that arrive at volume.

What defenders should watch for as the threat evolves

The most important shift is that normal user-experience controls take more pressure. Teams should expect more convincing initial contact, more variations of the same pretext and a greater chance that phishing will be combined with help desk impersonation, MFA fatigue, fake document sharing or other trust-based follow-on steps. The operational issue is less “can we spot one bad email?” and more “can we absorb many convincing attempts without a mistake propagating?”

Detection also becomes harder when campaigns are continuously varied. Signature-based filtering and static awareness examples lose value when the attacker can rotate wording, language and formatting at scale. Defences need to focus on verification points, behavioural anomalies and downstream compromise signals, not just obvious language errors or generic scam markers.

For a practitioner view of influence tactics, verification controls and awareness limitations, see Social Engineering Guide. For phishing variants, kits and defender countermeasures, see Phishing-as-a-Service Platforms Guide.

Risk and Threat Considerations

Malicious GenAI services increase phishing exposure by making it cheaper to run larger, more convincing campaigns across more targets. The threat is cumulative: even if any single lure is only moderately persuasive, the volume and personalisation can raise the chance of credential theft, account takeover or a successful secondary social-engineering step.

Failure mechanism: Attackers use GenAI to produce many context-aware lures quickly, then iterate on what gets responses, bypasses suspicion or triggers downstream action. That improves campaign efficiency without requiring a major increase in attacker skill.

Impact: Organisations face more frequent phishing pressure, more believable pretexts and a higher likelihood that routine controls, including user judgement and first-pass email filters, will be stressed at scale.

Practitioner Guidance

What to prioritise: Treat the problem as a scale-and-personalisation issue, not only a content-quality issue. If your control set only catches obvious scams, assume GenAI-enabled phishing will outpace it.

What to verify: Check whether your verification steps still hold when an email is tailored to a role, project or partner relationship. Any process that can be completed by replying to a message deserves an out-of-band check.

What good looks like: Users may still receive convincing lures, but they do not move sensitive requests forward without independent verification, and security teams can detect repeated campaign patterns even when the wording changes.

Practitioner takeaway: The defensive objective is to make phishing lures less actionable, not to assume they will become less believable; GenAI mostly changes attacker throughput and targeting precision, so controls must be resilient to volume and variation.

Related resources from NHI Mgmt Group

• Why do spear-phishing and malicious email campaigns remain such a serious risk in legal services environments?
• Why do malicious OAuth apps create more risk than a simple phishing email?
• Why do GenAI programmes create new identity risk even when the models change?
• How should security teams reduce malware risk from phishing and malicious downloads?