What Is Credential Theft?
Credential theft is the unauthorized acquisition of a user's authentication material: passwords, session cookies, OAuth tokens, MFA codes, API keys, or private keys. MITRE ATT&CK files it under the Credential Access tactic (opens in new tab) (TA0006), whose techniques are "for stealing credentials like account names and passwords."
Theft is the acquisition step itself: credential stuffing replays credentials stolen elsewhere, while account takeover, business email compromise, and ransomware deployment all sit downstream. Credential harvesting is one route to theft, the fake login page.
How credential theft works
Attackers take credentials at one of three points: while the user types them, while they sit at rest on a device, or while a session is live. Deceptive collection covers the first: a lookalike domain hosts a copy of a login page and the victim hands over the password.
Malware covers the second. An infostealer reads saved passwords and cookies out of the browser and bundles them into stealer logs. Interception covers the third: a reverse proxy sits between the victim and the real service and captures the authenticated session.
Tycoon 2FA shows interception at production scale. The phishing-as-a-service (PhaaS) kit emerged in August 2023 and sold through Telegram-based channels. It proxied the real Microsoft 365 login page: the victim entered a password and an MFA code, the kit passed both to the legitimate service, and the service returned an authenticated session cookie that the kit intercepted (opens in new tab).
The attacker replayed that cookie and entered the account with no second MFA prompt. Microsoft's Digital Crimes Unit, working with Europol and others, seized 330 domains behind the kit in March 2026.
Stolen material rarely stays with the thief. Infostealer operators sell logs to buyers that include initial access brokers, who resell enterprise footholds to ransomware affiliates. Mandiant's M-Trends 2026 (opens in new tab) puts the median hand-off from initial access to a secondary threat group at more than eight hours in 2022 and 22 seconds in 2025, because initial access partners now pre-stage the buyer's tooling at infection.
Why credential theft is hard to stop
A stolen credential produces a login that looks legitimate. The attacker authenticates as an authorized employee to the VPN, the cloud console, or the SaaS tenant, so perimeter controls have nothing to inspect. Cloudflare's 2026 threat report says identity abuse has superseded network exploitation (opens in new tab) as the primary vector for initial access, with attackers focused on "attacking the session" rather than "attacking the box."
MFA narrows the problem without closing it. CISA puts it plainly: "MFA solves some common attack vectors, like password guessing. But it doesn't solve all forms (opens in new tab) of credential theft." Strong authentication protects the login event; once the session cookie or refresh token exists, the thief inherits an already-authenticated session, the pattern behind session hijacking.
Password resets alone (opens in new tab) are insufficient, so responders must also revoke active sessions.
NYDFS 23 NYCRR Part 500 has generally required MFA for all users of covered systems since November 1, 2025, and a material credential-driven breach at a public company starts the SEC's four-business-day Form 8-K Item 1.05 clock.
Types of credential theft
Techniques group by where the attacker intercepts the credential.
- Fake login pages on lookalike domains. Credential harvesting clones a login page on a lookalike domain, often with AI-generated landing pages, and captures whatever the victim types.
- Adversary-in-the-middle (AiTM) proxy kits. Tycoon 2FA, EvilProxy, Sneaky2FA, and the open-source Evilginx framework relay the victim's session to the real service and keep the resulting cookie. The proxy defeats URL-reputation and visual checks because the victim interacts with the genuine page.
- Device code phishing. The attacker generates an OAuth device code and persuades the victim to approve it at the legitimate Microsoft login URL, which hands over access and refresh tokens without a password. PhaaS services such as EvilTokens and Kali365 have since packaged the technique.
- Infostealer malware. Lumma and Vidar harvest browser passwords, cookies, and wallet data. ClickFix lures, which trick users into pasting a command from a fake CAPTCHA into the Run dialog, deliver stealers on Windows, with Terminal-based variants on macOS.
- Vishing and help desk impersonation. Callers pose as IT staff or as the employee and talk the help desk into resetting a password or enrolling a new MFA device, a tactic NYDFS warned about (opens in new tab) in February 2026. Scattered Spider used help desk voice phishing to reset passwords and MFA tokens, per CISA (opens in new tab).
- MFA fatigue. An attacker who already holds the password floods the victim with push prompts until the victim approves one.
- Credential dumping (opens in new tab). Attackers with a foothold extract hashes and secrets from LSASS memory, the SAM database, and cached domain credentials. Password hashes then enable pass-the-hash lateral movement without the plaintext password.
How to defend against credential theft
Defense pairs stronger authentication with fast response and pressure on attacker infrastructure.
- Phishing-resistant MFA. FIDO2/WebAuthn credentials, the basis of passwordless authentication, bind the signature to the domain that registered them, so a proxy on a lookalike domain receives nothing usable. CISA calls phishing-resistant MFA the "gold standard for MFA" (opens in new tab) and ranks FIDO/WebAuthn and PKI above number-matching app tokens, with SMS and voice codes last.
- Token binding. Microsoft Entra token protection (opens in new tab) binds sign-in tokens to the issuing device for supported apps, and Google Workspace's Device Bound Session Credentials (opens in new tab) do the same for cookies in Chrome on Windows, so an exfiltrated token fails on attacker hardware. Pair binding with conditional access that forces phishing-resistant sign-in at elevated risk.
- A response that assumes the password is gone. When a leaked credential surfaces in an infostealer log or a dark web listing, reset the password, revoke active sessions and refresh tokens, remove unrecognized MFA registrations, and audit consented OAuth applications in one pass.
- Help desk verification. The help desk needs its own control, because verification for resets and MFA re-enrollment cannot rest on caller ID.
- External infrastructure monitoring. Monitor for lookalike domains and file takedown requests with registrars and hosts to remove the harvesting page, and enforce DMARC with SPF and DKIM so attackers cannot spoof your own domain in the lure.
- Training that mirrors live tactics. Simulation and training should run recurring exercises that include MFA fatigue prompts and help desk vishing calls.
Each control closes a gap the others leave open, so layer them.
How Doppel helps
Doppel is the Frontier AI Social Engineering Defense (SED) platform unifying Digital Risk Protection (DRP) and Human Risk Management (HRM). It targets credential theft where it starts: the spoofed login domain, the lure, and the employee it reaches.
The Doppel Threat Graph links the spoofed login domain, scam ad, vishing phone number, and leaked credential into one campaign, so teams do not investigate each signal in isolation. Brand Protection uses agentic AI to prioritize impersonation infrastructure and dismantle it through takedowns, with analysts validating high-confidence cases. Executive Protection monitors the dark web for leaked executive credentials so teams can act before the data is weaponized.
Dismantled harvesting pages also expose current lures, and the platform converts them into a live-threat Simulation in one click across email, SMS, voice, and Microsoft Teams. Phishing Triage handles reported phishing emails so analysts see only the cases that need a decision. Coordinated disruption of the domain, ad, and lure makes your brand too costly to attack.
Request a demo to get started.
Frequently asked questions about credential theft
What is credential theft?
Credential theft is an attacker obtaining the secrets a person or system uses to prove identity, including passwords, session cookies, OAuth tokens, API keys, and MFA codes. Attackers get them through fake login pages, reverse-proxy phishing kits, infostealer malware, calls to help desks, or a compromised machine's memory.
What is credential theft in cybersecurity?
Credential theft maps to the MITRE ATT&CK Credential Access tactic (opens in new tab) (TA0006), which groups techniques such as input capture, OS credential dumping, stealing web session cookies, MFA interception, and adversary-in-the-middle proxying. Together they frame theft as the acquisition step that precedes account takeover.
What is the difference between credential theft and credential stuffing?
Credential theft is the acquisition of credentials; credential stuffing is the automated reuse of credentials already stolen from one service against the login forms of others. Per OWASP (opens in new tab), stuffing "does not involve any brute-forcing or guessing of values", unlike password spraying, which tries a few common passwords across many accounts. Stuffing therefore depends on a prior theft, usually a breach or an infostealer log.
What is an example of credential theft?
Microsoft tracked Storm-2372, a group it assesses aligns with Russian interests, running device code phishing (opens in new tab) from August 2024 against government, defense, telecommunications, and energy organizations in Europe, North America, Africa, and the Middle East. Fake Microsoft Teams meeting invitations led targets to the legitimate device login page, where entering the attacker's code handed over valid access tokens. By February 2025 the group used the Microsoft Authentication Broker client ID to obtain refresh tokens and register attacker-controlled devices in Entra ID.

