WebAuthn 101

## Overview In this presentation, Christian Brand from Google’s security and identity teams addresses the fundamental limitations of password-based security and traditional multi-factor authentication (MFA). While password awareness and adoption of standards like FIDO have grown over the past five to six years (from nearly zero to roughly 40–50% audience familiarity), passwords remain insufficient against modern threats. Although Google's automated systems block unauthorized brute-force login attempts in over 99.9% of cases—proactively securing 110 million accounts by scanning leaked data—the talk focuses on securing the remaining 0.1%. At scale, this 0.1% accounts for high-value targets such as journalists, politicians, and corporate executives.

---

## The Scale and Mechanisms of Credential Theft Credential compromise is driven by three primary vectors: data breaches, password reuse, and phishing: - **Data Breaches & Reused Passwords:** Google search crawlers discovered over 4.3 billion unique leaked username/password combinations across public credential dumps. Cross-service password reuse occurs at a minimum rate of 17% (and likely much higher in practice), turning a single breach into an attack vector across multiple services. - **Phishing Dominance:** Phishing has become the single most dangerous and prevalent attack vector. In 2016, Google Safe Browsing data showed phishing surpassing exploit-based malware due to its low cost and high efficacy. - **Attack Statistics:** - 91% of information security attacks and 80% of business attacks initiate with or incorporate phishing. - A Google study showed that 43% of users (approximately 1 in 2) still fall for phishing websites when presented with urgent or legitimate-looking lures. - Accounts compromised via phishing are 463 times more likely to be hijacked than the average account. - According to Verizon’s Data Breach Investigations Report, weak or stolen passwords accounted for 76% of vulnerabilities (rising to 81% in subsequent reports). - **Secondary Vectors:** Traditional malware keyloggers and black-market credential brokers continue to exist, but phishing provides the lowest barrier to entry for active account hijacking.

---

## Limitations of Traditional Multi-Factor Authentication To counter password theft, the industry widely deployed MFA (such as SMS, voice calls, TOTP/HOTP generators, and mobile push notifications). While any MFA remains substantially better than passwords alone at preventing simple password-reuse attacks, standard MFA mechanisms remain inherently vulnerable to real-time relay attacks: - **Relay Attacks:** If an attacker directs a victim to a convincing reverse-proxy or phishing site, the attacker can submit the captured credentials to the legitimate service in real time. - **Bypassing Prompts:** When the real service issues an OTP prompt or mobile push verification, the user receives it, believes it corresponds to their own login attempt, and approves it or enters the code, directly authenticating the attacker.

---

## Evolution to FIDO, Security Keys, and WebAuthn To eliminate relayable authentication factors, industry collaborators formed the FIDO Alliance in 2012 and developed FIDO U2F (Universal 2nd Factor), the precursor to WebAuthn and FIDO2: - **Phishing-Resistant Design:** The core objective of the protocol is to bind authentication cryptographically to the specific origin/service, ensuring that credentials and signatures cannot be relayed or replayed by an attacker-controlled phishing site. - **Form Factors:** Although initially launched as dedicated physical USB/NFC hardware security keys for ease of hardware specification and manufacturing, the protocol is form-factor agnostic. - **Mobile Integration:** The underlying protocol has expanded beyond separate physical tokens to run directly on consumer devices, enabling smartphones to serve natively as phishing-resistant authenticators.