The AlphV ransomware group used stolen credentials to access Change Healthcare's portal. Attackers then moved laterally through the network, exfiltrated protected health information, and deployed ransomware that brought medical claims processing to a standstill. The breach ultimately affected 192.7 million individuals. All because compromised credentials bypassed a single unprotected entry point.
This attack shows how criminals use phishing to steal credentials, access business accounts, or install ransomware that locks systems and demands payment. According to the FBI's 2025 IC3 Annual Report, phishing and spoofing remained the most frequently reported cybercrime, while business email compromise (BEC) losses reached $3.046 billion.
Attackers favor email because employees trust familiar senders, common workflows, and everyday communication patterns. Legacy filters often struggle against these threats because they focus on static indicators rather than behavioral context. That gap helps explain why modern defenses need to evolve beyond signature-based detection.
Why Legacy Filters Miss Today's Malicious Email Attacks
Legacy filters often struggle with malicious email attacks that exploit trust rather than overt malicious code. Static, rule-based filters are less effective against threats built around impersonation, compromised accounts, and trusted platforms. Even sophisticated blocklists may not flag messages hosted on trusted cloud platforms or detect when legitimate vendor accounts send fraudulent payment requests.
Three gaps commonly drive this problem:
Static Indicators: Rules and signatures focus on known patterns rather than subtle changes in behavior.
Trusted Infrastructure: Messages sent through real accounts or reputable cloud services can appear legitimate at first glance.
Limited Context: A single-message inspection model may miss unusual timing, recipient patterns, or workflow deviations.
Modern detection requires behavioral context that understands how each sender normally writes, when they send, and to whom. The following twelve malicious email attack strategies show how cybercriminals evade traditional controls.
1. Trusted-Account BEC (Vendor/Partner Takeover)
Trusted-account BEC works because attackers send from real vendor accounts that already look legitimate.
Attackers who seize trusted vendor mailboxes request money, data, or goods in plain sight. From a supplier's compromised inbox, adversaries send from the real domain with perfect alignment across SPF, DKIM, and DMARC.
Teams can reduce risk by monitoring for changes that deviate from normal business patterns. Revised payment details that fall outside the usual workflow often signal trouble, as do out-of-cycle invoices that do not match the expected cadence. Requests arriving at unusual times compared with prior exchanges should also raise suspicion.
Secure email gateways (SEG) may detect little that looks unusual when language, signatures, and quoted threads mirror legitimate traffic. Abnormal's behavioral AI can help expose compromised partners through these shifts in vendor communication patterns.
2. Conversation Hijacking and Thread Injection
Conversation hijacking is effective because attackers insert themselves into existing threads that already carry trust. Attackers weaponize ongoing email threads to request money, credentials, or sensitive files. By compromising a mailbox, they gain access to conversation history and inject believable replies.
Common warning signs include:
Unverified Links: Replies that introduce unfamiliar domains or new login pages.
Tone Changes: Messages that sound different from the earlier thread.
Timing Shifts: Responses sent at unusual hours or outside the normal workflow.
Because messages appear within legitimate conversations, traditional trust indicators are less useful. Careful scrutiny can help expose the fraud before assets leave your organization.
3. AI-Written, Hyper-Personalized Social Engineering
AI-written phishing is harder to spot because attackers can generate polished, context-rich emails at scale. Attackers use large language models to craft personalized emails that can bypass both human skepticism and rule-based filters.
Generative AI removes many of the linguistic giveaways that once signaled danger, creating polished messages with accurate corporate jargon and details drawn from public profiles or breach data. OpenAI's own threat intelligence disclosures confirmed that named threat actors used ChatGPT to craft phishing emails, generate malicious scripts, and obfuscate HTML.
With each message slightly different, signature systems cannot easily catalog the variants. Abnormal's behavioral AI can help flag unusual tone, sequence, or urgency even in polished text.
4. Display-Name Spoofing
Display-name spoofing succeeds when recipients trust the name they see more than the address behind it. Display-name spoofing deceives recipients by pairing an executive's name with an unrelated legitimate domain that passes authentication checks.
This tactic exploits human psychology because users often scan sender names without scrutinizing full email addresses. Defending against it usually requires a combination of technical and procedural safeguards.
Deploying mailbox rules that flag name-domain mismatches is a useful starting point, paired with training that encourages employees to verify sender addresses before acting on urgent requests. Comparing sender history against established communication patterns adds another layer of confidence, and Abnormal's behavioral AI can help evaluate sender history against those patterns.
5. Supply-Chain Attacks via Compromised Partners
Supply-chain email attacks are dangerous because partner messages often look consistent with legitimate business activity. Attackers compromise supplier and partner inboxes to send malicious content that appears legitimate.
Messages from authentic domains can clear standard security checks, and teams may not question vendor requests that match existing purchase orders.
Helpful review points include:
Request Changes: Sudden shifts in what the partner is asking for.
Tone Changes: Wording that does not match prior messages.
Workflow Breaks: Requests that do not fit the normal purchasing process.
Establishing patterns for partner communication and flagging these changes can improve detection before a fraudulent invoice reaches the inbox.
6. QR-Code Phishing That Evades URL Scans
QR-code phishing can slip past inspection because the destination stays hidden until a user scans it.
Malicious QR codes embedded in image attachments bypass URL inspection by concealing destinations until scanned. These attacks redirect users to credential harvesting sites before email gateways detect suspicious traffic. Attackers often target mobile devices, where users may be less likely to inspect the underlying destination.
Effective protection blends user habits with technical controls. Encourage employees to preview QR destinations in a browser before signing in, and enforce multifactor authentication to reduce the impact of stolen credentials. Configuring email filters to flag image-only messages adds another layer of scrutiny, while Abnormal's behavioral AI can provide additional detection capability by analyzing attachment patterns against established user behavior.
7. Credential Harvesting via Legitimate Cloud Services
Credential harvesting through trusted cloud services is effective because the hosting domain often looks safe.
Attackers weaponize trusted platforms to steal credentials without triggering security alerts. Because these pages use trusted domains like Google Drive or SharePoint, reputation checks often clear them automatically.
Organizations can reduce risk through a few practical controls:
Phishing-Resistant MFA: Reduce the impact of stolen credentials.
Access Review: Review suspicious access patterns tied to identity activity.
Account Changes: Investigate unusual rule creation or suspicious OAuth grants.
Protection here depends on behavioral controls around suspicious access patterns and identity activity.
8. Time-Delayed, Multi-Stage Malicious Email Campaigns
Multi-stage campaigns work by spreading risk across messages so no single email looks clearly malicious.
Attackers sequence harmless and harmful messages over time, establishing credibility before launching attacks that bypass single-scan email controls. Traditional gateways evaluate messages upon arrival and then move on. Multi-stage campaigns exploit that limitation by using benign first touches and delayed follow-ups.
Several signals can hint at a campaign in progress. A sender may begin following an unusual communication rhythm, or later messages may sound noticeably different from the earlier outreach. New files or links appearing in connection with unfamiliar projects are another red flag. Detection requires analyzing relationship context across messages, and Abnormal's behavioral AI can help flag changes that follow a sequence of seemingly safe emails.
9. Invoice and Payment Fraud With Real PO Data
Invoice fraud is more convincing when attackers reuse real business context from compromised mailboxes. Attackers forge invoices that mirror genuine purchase orders to redirect payments before they are detected. They mine LinkedIn and compromised mailboxes for authentic PO numbers, line items, and approval chains.
Effective defense can include:
Secondary Verification: Verify payment changes through known contact information.
Transfer Delays: Consider waiting periods for high-value transfers.
Escalation Review: Treat banking-detail changes as potential breaches.
Abnormal's behavioral AI can help detect changes in vendor communication patterns surrounding these requests.
10. Domain and Look-Alike Spoofing
Look-alike spoofing relies on minor visual differences that recipients can miss at a glance.
Attackers register nearly identical domains using character substitutions or Unicode homographs to exploit user trust. Subtle variations like "rnicrosoft.com" using "rn" instead of "m" can appear legitimate during quick visual scanning.
A few practical measures help counter this tactic. Enabling punycode rendering in browsers exposes hidden Unicode tricks, and comparing sender domains against established communication patterns helps surface impostors. Requests tied to credential theft or financial changes deserve extra scrutiny, since attackers often pair domain trickery with high-stakes asks. Together, these controls can help identify deceptive attacks before users act on them.
11. Weaponized Docs and Links Hosted on Trusted Clouds
Trusted cloud hosting can give malicious documents and links the appearance of legitimacy.
Attackers conceal malware and phishing pages behind trusted platforms like Google Drive and Dropbox to bypass security controls. Legacy gateways rely on domain reputation, so links to trusted cloud services inherit the host's standing. Malicious payloads may activate one click deeper through HTML smuggling or JavaScript.
Useful review criteria include:
Sharing Patterns: Look for file-sharing behavior that does not match prior exchanges.
Authentication Prompts: Investigate unusual sign-in requests before opening content.
Attachment Context: Compare file size and message language against normal sharing habits.
Abnormal's behavioral AI can help flag unusual authentication demands before malicious content executes.
12. Internal Account Takeover and Lateral Phishing
Internal account takeover is dangerous because it turns a legitimate employee account into an attack channel.
Compromised employee mailboxes can spread phishing internally and externally while bypassing perimeter defenses. This makes it important to watch for suspicious access patterns and message bursts outside business hours.
Configuration changes can signal deeper compromise, and a layered response helps contain the damage. Implementing phishing-resistant MFA raises the bar for attackers, while rotating tokens when sessions appear suspicious cuts off active intrusions. Restricting mailbox permissions further reduces lateral movement during incidents, and Abnormal's behavioral AI can help quarantine emails that deviate from sender patterns.
Emerging Malicious Email Vectors: Deepfakes, AiTM, and Cross-Channel Attacks
Emerging malicious email vectors are becoming more effective as attackers combine inbox access with other channels and delivery methods. The twelve strategies above represent established attack patterns, but the threat landscape is accelerating in directions that compound their effectiveness.
Three trends stand out:
Deepfake Delivery: Email requests may be reinforced with voice or video that appears to validate the fraud.
AiTM Phishing: Proxy-based login pages can capture credentials and session cookies from a phishing email flow.
Cross-Channel Expansion: Email-borne social engineering can spill into collaboration tools where trust is already established.
Deepfake Audio and Video in BEC Delivery
Deepfake-assisted BEC extends email fraud by adding voice or video that appears to confirm the request.
Threat actors are pairing malicious email with deepfake audio and video to validate fraudulent email-initiated requests.
While these campaigns increasingly blend email with voice calls, text messages, and deepfake video, the primary control point remains the inbox. Abnormal's behavioral AI helps detect the email and account-based components of these scams, while organizations should pair this with additional controls for voice, SMS, and videoconferencing channels.
Adversary-in-the-Middle (AiTM) Session Hijacking
AiTM phishing adds session theft to credential theft, which raises the risk even when MFA is in place.
Phishing-as-a-Service (PhaaS) platforms now offer adversary-in-the-middle capabilities that intercept session tokens. These attacks start with a phishing email that routes victims through a proxy login page, capturing both credentials and session cookies. Because the initial email may contain only a link to a trusted-looking domain, URL-based filtering often struggles to intervene. Detection here depends on monitoring for session and behavioral signals that indicate compromised access.
Collaboration Platform Expansion
Email attacks increasingly spill into adjacent collaboration tools where trust is already established.
Attackers are extending email-borne social engineering into adjacent collaboration channels, where security visibility is often limited. Organizations need cross-channel behavioral coverage that extends beyond the inbox to protect cloud email and connected collaboration tools.
Abnormal's Behavioral AI: Closing the Malicious Email Gap
Abnormal helps close the malicious email gap by adding behavioral context to the email and account signals that legacy tools may miss.
Abnormal's behavioral AI transforms email security by establishing behavioral baselines through graph intelligence. The platform learns normal communication patterns within your organization, including workflow cadences, vendor interaction patterns, recipient behavior, and timing.
This approach is designed to detect subtle anomalies that traditional tools miss, such as wiring instructions sent outside business hours or sudden changes to banking details. Dynamic profiling continuously refines what constitutes usual behavior, helping surface threats like business email compromise or sophisticated phishing with fewer false positives.
Abnormal also strengthens existing defenses in practical ways:
Deployment: API-based deployment integrates with existing systems in minutes, with no MX record changes or endpoint agents.
Coverage: Abnormal extends protection across cloud email, Slack, Zoom, Teams, and Google Workspace using a unified intelligence engine.
Recognition: Recognized as a Leader in the Gartner® Magic Quadrant™ for Email Security Platforms.
Abnormal enhances the effectiveness of your current security stack while helping reduce false positives. Get a demo to see how Abnormal protects your organization against the full spectrum of evolving malicious email threats.
FAQ
What malicious email strategies bypass traditional security most often?
Trusted-account BEC, conversation hijacking, AI-written phishing, QR-code phishing, credential harvesting through legitimate cloud services, and internal account takeover are among the most effective because they exploit trust and behavior rather than obvious malicious payloads.
Why do legacy email filters miss these attacks?
Legacy filters rely heavily on static rules, signatures, domain reputation, and known indicators. Modern attacks often arrive from legitimate accounts, trusted domains, or personalized text that looks normal unless analyzed in behavioral context.
How can organizations reduce risk from malicious email attacks?
Use phishing-resistant MFA, verify payment or banking changes through secondary channels, monitor for login and mailbox anomalies, and deploy behavioral detection that understands normal sender-recipient patterns across email and collaboration tools.
What role does behavioral AI play in stopping malicious email?
Abnormal's behavioral AI helps identify subtle deviations in tone, timing, recipients, and transaction patterns that static tools miss, making it better suited to detect socially engineered threats, compromised accounts, and fraudulent requests hiding in legitimate-looking communication.