Abstract
The Internet is increasingly used to disseminate unethical and illegal content. A grave concern is child sexual abuse material that is often disseminated via end-to-end-encrypted channels. Such encryption defeats network- and server-based scanning measures used by law enforcement. A trade-off is to enable confidential communications channels for users and scanning opportunities for law enforcement by employing perceptual-hashing-based targeted content scanning on user devices. This has generated intense discussions between policymakers, privacy advocates and child protection organizations.
This chapter summarizes the current state of reserch in perceptual-hashing-based targeted content scanning with a focus on classical metrics such as false positives, false negatives and privacy aspects. Insights are provided into the most relevant perceptual hashing methods and an attack taxonomy for perceptual-hashing-based targeted content scanning is presented. The complexity in generating false negatives is evaluated and the feasibility of evading perceptual-hashing-based targeted content scanning is demonstrated.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
Similar content being viewed by others
References
H. Abelson, R. Anderson, S. Bellovin, J. Benaloh, M. Blaze, J. Callas, W. Diffie, S. Landau, P. Neumann, R. Rivest, J. Schiller, B. Schneier, V. Teague and C. Troncoso, Bugs in our pockets: The risks of client-side scanning, arXiv: 2110.07450v1 (arxiv.org/abs/2110.07450v1), 2021.
Apple, CSAM Detection – Technical Summary, Cupertino, California (www.apple.com/child-safety/pdf/CSAM_Detection_Technical_Summary.pdf), 2021.
Apple, Security Threat Model Review of Apple’s Child Safety Features, Cupertino, California (www.apple.com/child-safety/pdf/Security_Threat_Model_Review_of_Apple_Child_Safety_Features.pdf), 2021.
A. Athalye, Inverting PhotoDNA, Internet Archive, December 20, 2021.
A. Athalye, NeuralHash Collider, GitHub (github.com/anishathalye/neural-hash-collider), 2023.
A. Bhowmick, D. Boneh, S. Myers, K. Talwar and K. Tarbe, The Apple PSI System, Apple, Cupertino, California (www.apple.com/child-safety/pdf/Apple_PSI_System_Security_Protocol_and_Analysis.pdf), 2021.
F. Breitinger, B. Guttman, M. McCarrin, V. Roussev and D. White, Approximate Matching: Definition and Terminology, NIST Special Publication 800-168, National Institute of Standards and Technology, Gaithersburg, Maryland, 2014.
Facebook, Online child protection – Tools and technology, Internet Archive, June 21, 2022.
Federal Criminal Police Office, Presentation of the Numbers of Child Victims of Violence – Evaluation of the Police Crime Statistics 2021 (in German), Wiesbaden, Germany (www.bka.de/SharededDocs/Downloads/DE/AktuelleInformationen/Infografiken/Sonstige/kindlicheGewaltopfer_PKS2021.pdf), 2022.
Q. Hao, L. Luo, S. Jan and G. Wang, It’s not what it looks like: Manipulating perceptual-hashing-based applications, Proceedings of the ACM SIGSAC Conference on Computer and Communications Security, pp. 69–85, 2021.
S. Jain, A. Cretu and Y. Montjoye, Adversarial detection avoidance attacks: Evaluating the robustness of perceptual-hashing-based client-side scanning, Proceedings of the Thirty-First USENIX Security Symposium, pp. 2317–2334, 2022.
D. Kingma and J. Ba, Adam: A method for stochastic optimization, poster paper presented at the Third International Conference on Learning Representations, 2015.
E. Klinger and D. Starkweather, pHash: The open source perceptual hash library, Internet Archive, April 9, 2023.
N. Krawetz, Looks like it, Internet Archive, May 26, 2011.
N. Krawetz, Kind of like that, Internet Archive, January 21, 2013.
A. Kulshrestha and J. Mayer, Identifying harmful media in (end-to-end) encrypted communications: Efficient private membership computation, Proceedings of the Thirtieth USENIX Security Symposium, pp. 893–910, 2021.
Meta, Open-sourcing photo- and video-matching technology to make the Internet safer, Internet Archive, August 1, 2019.
Meta, The TMK+PDQF Video-Hashing Algorithm and the PDQ Image Hashing Algorithm, Menlo Park, California (raw.githubusercontent.com/facebook/ThreatExchange/main/hashing/hashing.pdf), 2021.
Microsoft, Photo DNA, Internet Archive, April 11, 2023.
National Center for Missing and Exploited Children, CyberTipline 2021 Report, Internet Archive, March 17, 2022.
D. Petrov, Wavelet image hash in Python, Internet Archive, July 2, 2016.
J. Prokos, T. Jois, N. Fendley, R. Schuster, M. Green, E. Tromer and Y. Cao, Squint hard enough: Evaluating perceptual hashing with machine learning, Cryptology ePrint Archive, paper no. 2021/1531 (eprint.iacr.org/2021/1531), 2021.
O. Russakovsky, J. Deng, H. Su, J. Krause, S. Satheesh, S. Ma, Z. Huang, A. Karpathy, A. Khosla, M. Bernstein, A. Berg and F. Li, ImageNet large scale visual recognition challenge, International Journal of Computer Vision, vol. 115(3), pp. 211–252, 2015.
L. Struppek, Learning-to-Break-Deep-Perceptual-Hashing, GitHub (github.com/ml-research/Learning-to-Break-Deep-Perceptual-Hashing), 2022.
L. Struppek, D. Hintersdorf, D. Neider and K. Kersting, Learning to break deep perceptual hashing: The use case NeuralHash, Proceedings of the ACM Conference on Fairness, Accountability and Transparency, pp. 58–69, 2022.
K. Walker, Four steps we’re taking today to fight terrorism online, Internet Archive, June 18, 2017.
Z. Wang, A. Bovik, H. Sheikh and E. Simoncelli, Image quality assessment: From error visibility to structural similarity, IEEE Transactions on Image Processing, vol. 13(4), pp. 600–612, 2004.
WeProtect Global Alliance, Global Threat Assessment 2021, Sevenoaks, United Kingdom (www.weprotect.org/wp-content/uploads/Global-Threat-Assessment-2021.pdf), 2021.
B. Yang, F. Gu and X. Niu, Block mean value based image perceptual hashing, Proceedings of the International Conference on Intelligent Information Hiding and Multimedia, pp. 167–172, 2006.
C. Zauner, Implementation and Benchmarking of Perceptual Image Hash Functions, M.S. Thesis, Secure Information Systems Program, University of Applied Sciences Upper Austria, Hagenberg, Austria, 2010.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2023 IFIP International Federation for Information Processing
About this chapter
Cite this chapter
Twenning, L., Baier, H., Göbel, T. (2023). Using Perceptual Hashing for Targeted Content Scanning. In: Peterson, G., Shenoi, S. (eds) Advances in Digital Forensics XIX. DigitalForensics 2023. IFIP Advances in Information and Communication Technology, vol 687. Springer, Cham. https://doi.org/10.1007/978-3-031-42991-0_7
Download citation
DOI: https://doi.org/10.1007/978-3-031-42991-0_7
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-42990-3
Online ISBN: 978-3-031-42991-0
eBook Packages: Computer ScienceComputer Science (R0)