Phishing, AI and Patients: New Risks as Email Gets Smarter

When your inbox gets smarter, the scams do too — and patients pay the price

Hook: If you use email for appointment reminders, lab results or billing notices, the new wave of AI-powered inbox features introduced in late 2025 and early 2026 changes the threat landscape. Smarter Gmail summaries, automatic sorting and AI-generated overviews can save time — but they also make it easier for targeted phishing to look legitimate. Patients and caregivers must adapt their digital hygiene now or risk identity theft, medical fraud, and compromised care continuity.

The landscape in 2026: why email AI matters for patient safety

Beginning in late 2025 and continuing through early 2026, major email platforms rolled out inbox capabilities powered by large multimodal models (Google’s Gemini family is a prominent example). These features include AI-generated summaries, suggested replies, auto-categorization of messages, and smarter attention flags. For busy patients and caregivers, these innovations can be a time-saver — but they also give attackers new levers for social engineering.

How attackers are adapting

• AI-written messages: Phishers now use the same models that power inbox summaries to craft highly personalized, typo-free messages that match tone and context.
• Summary baiting: A malicious email may contain a short, convincing lead that an AI summary highlights, increasing the chance a user will act on the summary without opening the full message.
• Targeted data fusion: Attackers combine leaked personal data (names, providers, appointment dates) with AI to produce messages that exploit trust — for example, an email that appears to reference yesterday’s lab draw.
• Filter evasion: Small changes to subject lines and message structure — informed by testing and AI — can help phishing messages slip past automated spam filters and into primary inboxes.
• Silent forwarding and account takeover: Once access is obtained, attackers can create forwarding rules or bulk-delete messages to cover tracks; smarter inbox UIs can hide these configurations from casual review.

Why patients and caregivers are high-value targets

Health-related emails often contain actionable requests (confirm an appointment, download lab results, pay a bill) and personally identifiable information. That combination makes patients attractive to criminals looking to commit medical identity theft, insurance fraud, or credential harvesting for larger campaigns. Patients should ask clinics about secure intake and delegation features that reduce risky password-sharing.

Real-world scenarios: two short case studies

Case study 1 — The lab result summary

Maria, 62, received an AI-generated overview prompting “Abnormal lab result—see attached.” Her Gmail preview displayed a concise summary that looked like a legitimate alert from her clinic. She clicked the link, entered her patient portal credentials on a convincing fake site and later learned her insurance benefits were changed without consent. A quick check of the full message and a phone call to the clinic would have exposed the fake sender domain; providers that publish clear verification guidance and delegation workflows (see provider intake best practices) make these checks easier.

Case study 2 — The caregiver inbox takeover

Jamal manages his mother’s care and shares her provider’s emailed communications in a delegated inbox. An attacker used a convincing billing email, got Jamal to click a link, and set a forwarding rule to a burner address. The attacker then arranged fraudulent prescriptions. Jamal had shared one password across multiple sites; when he changed it and enabled passkeys and stronger MFA, the malicious forwarding was discovered and removed.

Key principles for protecting patients in an AI-driven inbox era

Three simple rules can reduce risk immediately:

1. Don’t act on summaries alone. Open the full email and verify sender details before clicking links or sharing credentials. If in doubt, use privacy-first browsing practices (see privacy-first browsing) and check the original message.
2. Verify via a trusted channel. If an email asks for action (pay, reset password, download results), call the provider using a number from a recent statement or their official website — not one inside the email.
3. Lock the account down. Use multi-factor authentication (MFA), a password manager, and periodic security checks. For large-scale guidance on rotation and MFA best practices see Password Hygiene at Scale.

Practical, step-by-step actions patients and caregivers should take today

The recommendations below are practical and prioritized: immediate actions to reduce risk, followed by defensive habits and settings to review.

Immediate checklist (what to do right now)

• Stop relying on AI summaries: When a message references health data (results, billing, prescriptions), open the full message before acting. Summaries can omit malicious details or elevate malicious text.
• Verify the sender domain: Hover over the sender name to reveal the email address. Confirm the domain matches the clinic/insurer (e.g., @exampleclinic.org). Beware of close typos or extra words.
• Don’t click embedded links for urgent requests: Instead, open a browser and type the provider’s known URL or use your patient portal app.
• Check for unknown forwarding rules: In Gmail, go to Settings > Forwarding and POP/IMAP and ensure no unknown addresses are receiving your mail. If you find suspicious forwarding, follow steps from an incident response template to contain the compromise.
• Run a security checkup: Use Google Account > Security Checkup (or your mail provider’s equivalent) to review connected devices and apps, and remove anything you don’t recognize. See large-scale password and MFA guidance to harden accounts.

Configure your inbox for safety

Take these configuration steps in Gmail and most major providers. If you use another client, look for equivalent settings.

• Review images and external content: In Gmail settings, set images to “Ask before displaying external images.” External images can confirm your address to attackers and automatically load malicious resources in some cases — privacy-focused browsing guidance is helpful here (privacy-first browsing).
• Inspect filters and blocked addresses: Go to Settings > See all settings > Filters and Blocked Addresses and remove any unfamiliar rules. Attackers often add filters to hide or auto-archive messages.
• Enable multi-factor authentication (MFA): Use an authenticator app or security key (FIDO2) instead of SMS when possible. In 2026, passkeys and hardware keys are widely supported and far more resistant to phishing — see password hygiene best practices.
• Use unique, strong passwords: Store them in a reputable password manager and avoid reusing passwords across healthcare portals and other accounts.
• Limit automatic account access: Revoke third-party apps with broad scopes you don’t need. In Gmail, check Security > Third-party apps with account access; for vendor transparency and auditability expect providers to support edge auditability and decision planes.

Caregiver-specific controls

Caregivers manage multiple accounts and shared access, creating specific risks. Apply these controls:

• Use delegated access, not shared passwords: Where available, ask clinics for delegation features or official caregiver links instead of sharing credentials. Provider intake and delegation guidance is covered in resources on advanced patient intake.
• Create separate contact lists and aliases: Use a dedicated email alias for health communications (or Gmail plus-addressing) so you can quickly filter messages and detect unexpected senders.
• Set up alerts for sign-in activity: Enable notifications for new device sign-ins and suspicious activity. Agree with the care recipient on how these alerts will be handled — if a compromise is suspected, use an incident response checklist to act quickly.
• Limit device privileges: Avoid storing patient credentials on devices used by multiple people. Use the “guest” or separate user accounts for shared devices.

Recognizing the new signs of AI-enhanced phishing

Attackers exploit the very features designed to help. These are the new red flags to watch for:

• Overly concise but contextually specific summaries: If a preview or AI summary mentions a specific procedure, date or test result but the full email is vague, treat it as suspicious — an example of why AI should augment, not replace, human checks.
• Misspelled or mismatched domain names: Tiny differences in a sender domain (rnstead of .org) are common tricks.
• Unexpected attachments or urgency: Any email pressuring you to click a link or download an attachment right away should be verified independently.
• Messages routed to ‘Primary’ despite not being from contacts: AI categorization models may put crafted phishing into primary; double-check sender identity.

Technical checks for the curious and capable

For patients or caregivers comfortable with deeper technical checks, these steps help verify message authenticity.

• View original headers: In Gmail, choose More > Show original. Check SPF, DKIM and DMARC authentication results — look for pass/fail indicators. Vendor transparency and auditability frameworks can help providers publish authoritative verification guidance (edge auditability & decision planes).
• Search for forwarding rules or filters added recently: Attackers often create filters that auto-delete or archive notifications to hide their activity.
• Audit connected apps: Revoke OAuth tokens for unfamiliar apps; attackers sometimes gain access via consent grants to malicious apps — see large-scale password and app-access guidance.

What to do if you were phished or think your account is compromised

Quick containment limits damage. Follow this recovery checklist immediately:

1. Disconnect and don’t use the account: Log out of all devices from your account’s security page.
2. Change the password and enable MFA: Use a new, unique password and preferably a passkey or authenticator app.
3. Check forwarding and filters: Remove unknown rules and blocked addresses.
4. Revoke third-party app access: Remove any OAuth permissions that look unfamiliar.
5. Contact affected providers: Call your clinic, insurer or pharmacy using a trusted phone number to report suspicious activity and confirm account changes.
6. Monitor accounts and credit: Consider an identity protection service or credit freeze if sensitive data was exposed.
7. Report the phishing: In Gmail click More > Report phishing or forward the message to your provider’s security or support contact. Reporting helps block similar attacks; follow an incident response template for next steps.

What healthcare organizations and platforms must do (brief overview for patients to ask about)

Patients and caregivers can pressure providers to adopt safer practices. Ask your clinic or insurer whether they:

• Use secure patient portals for sending sensitive results instead of unencrypted email; see examples in advanced patient intake documentation;
• Publish and verify official email domains and shortcodes so patients can check authenticity; this is easier when vendors support auditability and transparent decision planes;
• Support delegation and caregiver features that avoid password-sharing; delegation workflow advice appears in provider intake guidance;
• Enroll patients in account security checks and provide clear instructions for reporting suspicious messages — a simple incident checklist like this incident response template helps staff respond quickly.

Trends and predictions for the next 24 months (2026–2028)

Looking ahead, these developments will shape the threat and defense landscape for patient email safety:

• More inbox AI, more nuanced attacks: As providers and criminals both use advanced models, expect attackers to produce higher-fidelity impersonations. The arms race will continue.
• Passkeys and hardware-backed auth become standard: Adoption of phish-resistant authentication (passkeys, security keys) will rise, especially where regulators incentivize stronger controls for health data — follow password hygiene guidance.
• Regulatory focus on AI and PHI: Health regulators will further clarify obligations for vendors that process protected health information with AI models — expect more vendor due-diligence and transparency requirements; resources on edge auditability are relevant.
• Stronger native protections in email clients: Providers will add protections like “suspicious content” banners powered by in-line verification models and easier opt-out from AI summarization for sensitive mail categories; look for native privacy and edge-host solutions such as pocket edge hosts that make inline verification simpler.

Bottom line: practical patient-safety behaviors for an AI inbox world

AI features in email bring convenience but also new phishing risk vectors. Protecting patients requires a mix of common-sense behaviors, secure account configuration, and coordination with healthcare providers. The most effective habits are simple and repeatable:

• Always verify. Use a trusted channel for confirmation when messages request action.
• Harden accounts. Enable MFA/passkeys, unique passwords, and regular security reviews — see password hygiene at scale.
• Control the mailbox. Regularly inspect forwarding rules, filters and app access.
• Prefer patient portals. When possible, receive lab results and billing through secure portals rather than email attachments or links; providers should publish clear intake and delegation guidance (advanced patient intake).

“Smarter inboxes help, but never replace the single human check: verify before you act.”

Quick printable checklist (summary)

• Don’t rely on AI summaries — open the full email.
• Hover to view sender email and confirm the domain.
• Never click links in urgent-sounding health emails; call the provider.
• Enable MFA/passkeys and use a password manager.
• Check Gmail: Forwarding & POP/IMAP, Filters & Blocked Addresses, Security Checkup.
• Use patient portals for sensitive data whenever possible.
• Report phishing to your provider and mail provider immediately (see an incident response template).

Final thoughts and a clear next step

In 2026 the mailbox is smarter — and so are the threats. Patients and caregivers who adopt a few defensive habits will stay safer without sacrificing convenience. Start by running a security checkup on your email account and enabling passkeys or an authenticator app. Teach these steps to anyone who helps manage care; a small change can prevent major harms.

Call-to-action

Take 10 minutes now: run your account’s Security Checkup, review Forwarding and Filters, and enable multi-factor authentication. Share this article with the family member or caregiver who handles medical messages. If you want a printable checklist tailored for caregivers, visit themedical.cloud/security-checklist to download one and sign up for our monthly patient-safety brief.

Related Reading

• Incident Response Template for Document Compromise and Cloud Outages
• Password Hygiene at Scale: Automated Rotation, Detection, and MFA
• Edge Auditability & Decision Planes: An Operational Playbook for Cloud Teams in 2026
• Advanced Patient Intake: Trauma-Informed Workflows for Clinics
• Build a Landing Page for Social Spikes: Capture Traffic When a Platform Goes Viral
• How to Harden Client Communications and Incident Response for Studios (2026 Checklist)
• The Best TSA-Friendly Beauty Organizers for New Product Launches
• DIY Cocktail Syrup: Make Your Own Tiny-Batch Syrups for Gifts
• Mini‑Me, Pooch Edition: How to Nail Matching Outfits with Your Dog