Evaluating the Security Posture of Real-World FIDO2 Deployments
Open accessThe DOI matches its Crossref record; 38 of the 38 references it cites were found in a registry or an index; 7 of the 60 sources its claims cite were checked for support.
Abstract
FIDO2 is a suite of protocols that combines the usability of local authentication (e.g., biometrics) with the security of public-key cryptography to deliver passwordless authentication. It eliminates shared authentication secrets (i.e., passwords, which could be leaked or phished) and provides strong security guarantees assuming the benign behavior of the client-side protocol components.
Overview
CoChat Summary
## TL;DR
This paper studies how real-world FIDO2 deployments handle compromised clients, using measurements from 29 relying parties, a year-plus study of the FIDO Metadata Service, and telemetry from a large financial-services RP. It finds that many deployments do not adopt recommended mitigations, that most authenticators lack malware-resistance certification, and that practical malware and social-engineering attacks remain feasible.
## Key findings
- The authors evaluated 29 RPs from the Tranco Top 1K: 85 sites in the Top 1K supported WebAuthn, collapsing to 40 distinct RPs, of which 29 were evaluable.
- In the MDS snapshot from April 13, 2023, there were 160 authenticators; only 7/160 (4%) were certified L2 (malware resistant), while 120 were L1 and 58 supported no user verification at all.
- Among the 29 RPs, only 14/29 requested any attestation, 27/29 allowed a virtual authenticator, and only 3/29 required user verification; only 5/29 declared a preference for platform authenticators.
- From partner-RP telemetry, the authors observed ∼72% Apple-attested, ∼27% Android SafetyNet-attested, and ∼1% unattested WebAuthn responses; they also report evidence of system integrity compromise on ∼0.25% of hardware-backed clients and ∼13% of non-hardware-backed clients.
## Methods
- The study combines three measurements: (i) manual/semi-automated analysis of FIDO2 configurations on top websites, (ii) a longitudinal MDS study from September 16, 2021 to April 13, 2023, and (iii) real-world authenticator telemetry from a large financial-services RP.
- For RP measurement, the authors manually created accounts and inspected WebAuthn registration flows on candidate sites, then tested authenticator registration with Chrome’s virtual authenticator under different transports and protocol settings.
- For telemetry, they analyzed 126,608 payloads on July 20, 2022 and 74,270 on August 23, 2022 from the partner RP, plus a 14,616-session validation sample with full payloads.
## Limitations
- The paper states that the partner-RP telemetry was limited to mobile browser sessions and thus the real-world authenticator sample is restricted to mobile devices.
- It also states that server logging only captured the first 4KB of each payload, so some responses were truncated and not parseable.
- Evident from the text: the RP configuration study is limited to 29 RPs from the Tranco Top 1K and may not generalize to all FIDO2 deployments.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Figures



Verification
Claims the paper makes
2 Valid · 13 Not valid · 44 Not checked · 147 Own results and interpretations
The cited sources that could be checked couldn’t confirm or dispute this claim.
In the paper
“when this assumption does not hold true, such as in the presence of malware, client authentications pose a risk that FIDO2 deployments must account for.”
- Also stated in: § 3 RELATED WORK · hedged
“these properties do not necessarily hold with malicious client components.”
§ Evaluating the Security Posture of Real-World FIDO2 Deployments > Abstract
What the sources say
- Not matchedcites [56] 56. Namecheap Knowledgebase. 2022. How can I enable/disable Two-Factor Authenti- cation? https://www.namecheap.com/support/knowledgebase/article.aspx/9253/ 45/how-can-i-enabledisable-twofactor-authentication/.
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Could not verifycites [57] 57. Takashi Koide, Daiki Chiba, and Mitsuaki Akiyama. 2020. To Get Lost is to Learn the Way: Automatically Collecting Multi-step Social Engineering Attacks on the Web. In ACM Asia Conference on Computer and Communications Security (AsiaCCS). sourceExperimental — accuracy not yet measured
The Background and PUP and Malware sections discuss web-based social-engineering attacks and malware downloads. Searches for FIDO2, client authentication, and compromised client-side components found no passage addressing the claimed risk to FIDO2 deployments. (LLM, conf=0.78)
plainstates a possible cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Also stated in: § 3 RELATED WORK · hedged
The cited sources that could be checked couldn’t confirm or dispute this claim.
In the paper
“This threat model is realistic as in practice, keyloggers have been widely used to steal user passwords [78].”
§ Evaluating the Security Posture of Real-World FIDO2 Deployments > 4 THREAT MODEL
What the sources say
- Could not verifycites [78] 78. Stephan Wiefling, Luigi Lo Iacono, and Markus Dürmuth. 2019. Is This Really You? An Empirical Study on Risk-Based Authentication Applied in the Wild. In IFIP International Conference on ICT Systems Security and Privacy Protection. Springer. sourceExperimental — accuracy not yet measured
Our copy of Wiefling et al. (2020) never mentions keyloggers: searches for 'keylogger', 'stealing passwords', and 'malware/keystroke logging' returned only the attacker-model section (2.2 Attacker Models, p. 5), which defines three attackers — password guesser, credential-stuffing attacker (from database leaks), and phishing/MITM attacker — and the Introduction (p. 1), which names large-scale password database leaks and credential stuffing as the growing threats. The paper's model explicitly assumes 'the attacker does not have physical access or eye contact with the victim's devices' (p. 5), so it neither reports nor relies on keylogger use; nothing in it contradicts the claim either, so the attribution cannot be checked against this source. (LLM, conf=0.70)
boostedstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Could not verify
The cited sources that could be checked couldn’t confirm or dispute this claim.
In the paper
“provides strong security guarantees assuming the benign behavior of the client-side protocol components.”
- Also stated in: § 1 INTRODUCTION · boosted
“FIDO2 has been designed and formally verified [30,32,56,57] as providing strong authentication security guarantees”
§ Evaluating the Security Posture of Real-World FIDO2 Deployments > Abstract
What the sources say
- Not checked — no full text of the sourcecites [30] 30. Nina Bindel, Cas Cremers, and Mang Zhao. 2023. FIDO2, CTAP 2.1, and WebAuthn 2: Provable security and post-quantum instantiation. In IEEE Symposium on Security and Privacy (IEEE S&P). source
- Could not verifycites [32] 32. David Cerdeira, Nuno Santos, Pedro Fonseca, and Sandro Pinto. 2020. SoK: Understanding the Prevailing Security Vulnerabilities in TrustZone-assisted TEE Systems. In IEEE Symposium on Security and Privacy (S&P). sourceExperimental — accuracy not yet measured
The paper’s D04 and I08 sections discuss TrustZone-assisted TEEs, secure boot, and remote attestation, not FIDO2’s authentication guarantees. Searches for FIDO2, FIDO client authenticators, and benign client-side protocol components found no passage addressing the attributed claim. (LLM, conf=0.77)
- Not matchedcites [56] 56. Namecheap Knowledgebase. 2022. How can I enable/disable Two-Factor Authenti- cation? https://www.namecheap.com/support/knowledgebase/article.aspx/9253/ 45/how-can-i-enabledisable-twofactor-authentication/.
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Could not verifycites [57] 57. Takashi Koide, Daiki Chiba, and Mitsuaki Akiyama. 2020. To Get Lost is to Learn the Way: Automatically Collecting Multi-step Social Engineering Attacks on the Web. In ACM Asia Conference on Computer and Communications Security (AsiaCCS). sourceExperimental — accuracy not yet measured
The Conclusion describes StraySheep’s detection of multi-step social-engineering attacks, while §3.2.2 discusses browser navigation and redirection. Searches for FIDO2 and client-side authentication guarantees found no passage addressing the claimed security assumption. (LLM, conf=0.76)
boostedstates a possible cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Also stated in: § 1 INTRODUCTION · boosted
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“In total, we aggregated the FIDO2-supporting domains into 40 distinct RPs.”
§ Evaluating the Security Posture of Real-World FIDO2 Deployments > 5.1.1 Identifying RPs deploying FIDO2.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“We did not observe any other attestations.”
§ Evaluating the Security Posture of Real-World FIDO2 Deployments > Attestation Type Apple Anonymous None
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
77 more claims
Create a free account to see all 206 claims and to verify the 21 references still open.
- It may state two separate claims as one. 79
- Probably not a substantial claim. 66
- The quote only partly states this claim. 1
Citations
Cites
(2011) · IEEE Software · doi:10.1109/ms.2011.67 · Cited by 1,989
2021 IEEE Symposium on Security and Privacy (SP)
(2021) · doi:10.1109/sp40001.2021 · Cited by 286
SoK: Understanding the Prevailing Security Vulnerabilities in TrustZone-assisted TEE Systems
Cerdeira, D., Santos, N., Fonseca, P., & Pinto, S. (2020) · 2020 IEEE Symposium on Security and Privacy (SP) · doi:10.1109/sp40000.2020.00061 · Cited by 208 · cited 1×
“As an increasing number of devices and services bank upon the security guarantees provided by TEEs, other TEE implementations have also been similarly targeted [33, 34].”
Background(2017) · doi:10.1145/3133956.3134067 · Cited by 187 · cited 2×
“This threat model is realistic as in practice, keyloggers have been widely used to steal user passwords [78].”
BackgroundMethodSelf-citationSoK: The Challenges, Pitfalls, and Perils of Using Hardware Performance Counters for Security
Das, S., Werner, J., Antonakakis, M., Polychronakis, M., & Monrose, F. (2019) · 2019 IEEE Symposium on Security and Privacy (SP) · doi:10.1109/sp.2019.00021 · Cited by 179 · cited 1×
“The FIDO2 Client’s privileged access could further be utilized to fingerprint the user’s presence via kernel-level events [28], device characteristics [29, 31], or even hardware-level features [35].”
BackgroundDevice fingerprinting for augmenting web authentication: classification and analysis of methods
Alaca, F., & van Oorschot, P. C. (2016) · Annual Computer Security Applications Conference · doi:10.1145/2991079.2991091 · Cited by 145 · cited 1×
“The FIDO2 Client’s privileged access could further be utilized to fingerprint the user’s presence via kernel-level events [28], device characteristics [29, 31], or even hardware-level features [35].”
BackgroundGhorbani Lyastani, S., Schilling, M., Neumayr, M., Backes, M., & Bugiel, S. (2020) · 2020 IEEE Symposium on Security and Privacy (SP) · doi:10.1109/sp40000.2020.00047 · Cited by 128
Who Are You? A Statistical Approach to Measuring User Authenticity
Freeman, D., Jain, S., Duermuth, M., Biggio, B., & Giacinto, G. (2016) · Proceedings 2016 Network and Distributed System Security Symposium · doi:10.14722/ndss.2016.23240 · Cited by 123 · cited 2×
“We propose that the security community investigate augmenting existing authentication risk models [51, 63, 82, 84] to include additional risk signals on whether a FIDO2 authentication attempt was initiated by the intended user.”
BackgroundUpgrading Your Android, Elevating My Malware: Privilege Escalation through Mobile OS Updating
Xing, L., Pan, X., Wang, R., Yuan, K., & Wang, X. (2014) · 2014 IEEE Symposium on Security and Privacy · doi:10.1109/sp.2014.32 · Cited by 111 · cited 1×
“We consider both a weaker threat model of malware with only user-level access (e.g., a malicious user application or browser extension) and a stronger threat model of malware that achieves elevated root access (such as through exploiting kernel vulnerabilities [86]).”
MethodIs This Really You? An Empirical Study on Risk-Based Authentication Applied in the Wild
Wiefling, S., Iacono, L. L., & Dürmuth, M. (2019) · IFIP advances in information and communication technology · doi:10.1007/978-3-030-22312-0_10 · Cited by 95
Research works printed but not linked: 7
- Florian M Farke, Lennart Lorenz, Theodor Schnitzler, Philipp Markert, and Markus Dürmuth. 2020. You still use the password after all - Exploring FIDO2 Security Keys in a Small Company. In Symposium on Usable Privacy and Security (SOUPS).
- Anthony Gavazzi, Ryan Williams, Engin Kirda, Long Lu, Andre King, Andy Davis, and Tim Leek. 2023. A Study of Multi-Factor and Risk-Based Authentication Availability. In USENIX Security Symposium (USENIX Security).
- Bank of America: Privacy and Security. 2022. Secured Transfer. https://www.ba nkofamerica.com/security-center/faq/additional-security-features/.
- Kentrell Owens, Olabode Anise, Amanda Krauss, and Blase Ur. 2021. User perceptions of the usability and security of smartphones as FIDO2 roaming authenticators. In Symposium on Usable Privacy and Security (SOUPS).
- Alon Shakevsky, Eyal Ronen, and Avishai Wool. 2022. Trust Dies in Darkness: Shedding Light on Samsung's TrustZone Keymaster Design. In USENIX Security Symposium.
- Roman Unuchek. 2017. Kaspersky Daily: Rooting your Android. https://usa.kasp ersky.com/blog/android-root-faq/11581/.
- Tara Whalen, Thibault Meunier, Mrudula Kodali, Alex Davidson, Marwan Fayed, Armando Faz-Hernández, Watson Ladd, Deepak Maram, Nick Sullivan, Benedikt Christoph Wolters, et al. 2022. Let The Right One In: Attestation as a Usable CAPTCHA Alternative. In Symposium on Usable Privacy and Security (SOUPS).
Cited by
How many FIDO protocols are needed? Analysing the technology, security and compliance
Angelogianni, A., Politis, I., & Xenakis, C. (2024) · ACM Computing Surveys · doi:10.1145/3654661 · Cited by 16
Yang, J.-P., Fang, B., Lu, H. D., & Tian, Z. (2024) · IEEE Internet of Things Journal · doi:10.1109/jiot.2024.3515079 · Cited by 8
“However, Kuchhal et al. [51] identified multiple security risks in FIDO2 implementation.”
Holtgrave, J.-U., Klivan, S., Marky, K., & Fahl, S. (2025) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3706598.3713252 · Cited by 11
“There have been some small security issues related to FIDO2-based authentication [39], but the overall security claims are still withheld.”
Matzen, A., Rüffer, A., Byllemos, M., Heine, O., Παπαϊωάννου, Μ., Choudhary, G., & Dragoni, N. (2025) · Applied Sciences · doi:10.3390/app15084414 · Cited by 13
Mahdad, A. T., Jubur, M., & Saxena, N. (2024) · ACM SIGSAC Conference on Computer and Communications Security (CCS) · doi:10.1145/3658644.3690286 · Cited by 10
“Kucchal et al. [28] conducted an evaluation of real-world FIDO2 deployments, demonstrating how malware in the user’s terminal can bypass FIDO2 authentication attempts.”
Influential
Metadata
- Authors
- Kuchhal, D., Saad, M., Oest, A., & Li, F.
- Affiliations
- Georgia Institute of Technology (US)
- PayPal (United States) (US)
- Venue
- Proceedings of the 2023 ACM SIGSAC Conference on Computer and Communications Security
- Publisher
- ACM
- Published
- 2023-11-15
- Pages
- 2381-2395
- ISSN
- 1543-7221 (ISSN-L)
- OpenAlex
- W4388858368
- Cited by
- 22
- References
- 26
- Citations by year
- 2026: 8 · 2025: 8 · 2024: 6
- FWCI
- 5.64
- Type
- conference paper
- Language
- en
- Open access
- hybrid
- doi.org
- Locations
- ACM SIGSAC Conference on Computer and Communications Security (CCS) (publishedVersion, cc-by)
- Funders
- National Science Foundation (CNS-2055549)
- Topics
- Physical Sciences › Computer Science › Information Systems › User Authentication and Security Systems
- Physical Sciences › Computer Science › Computer Networks and Communications › Advanced Authentication Protocols Security
- Physical Sciences › Computer Science › Hardware and Architecture › Security and Verification in Computing
- Keywords
- FIDO2, passwordless authentication, public-key cryptography, biometric authentication, authentication protocol security
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol and qwen3.8-omni-flash).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Evaluating the Security Posture of Real-World FIDO2 Deployments. https://app.cochat.ai/papers/10.1145/3576915.3623063
- The paper
- Kuchhal, D., Saad, M., Oest, A., & Li, F. (2023). Evaluating the Security Posture of Real-World FIDO2 Deployments. Proceedings of the 2023 ACM SIGSAC Conference on Computer and Communications Security. https://doi.org/10.1145/3576915.3623063
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.