2024-02-28 DeepfakesIDV DG Meeting Notes DRAFT

Ajay Amlani

iProov

Andrew Hughes

FaceTec, Inc.

Present

Ashok Singal

Demystify Biometrics

Chris LaBarbera

Verizon

Denny Prvu

RBC Royal Bank of Canada

Iain Corby

SafetyTech Limited

Jay Meier

FaceTec, Inc.

Present

Jim Pasquale

Chasm Crossing Solutions

Present

Jordan Burris

Socure

Justin Byrd

N/A

Present

Maxine Most

Acuity Market Intelligence

Present

Mike Magrath

Easy Dynamics

Present

Peter Davis

Airside Mobile, Inc

Present

Sean Lanzner

iProov

Present

Terry Brenner

FIDO Alliance

Anand Kumar

SecureKloud Technologies, Inc.

Present

Dan Bachenheimer

Accenture

Present

Dawid Jacobs

DAL Identity

Present

Michael Choudoin

N/A

Present

Michael Engle

1Kosmos

Present

Pieter Van Iperen

N/A

Present

Stephanie Schuckers

Clarkson University

Present

Uttam Reddi

Aware Corp

Present

Kay Chopard

Kantara

Present

Biometric inputs

Mike Chaudoin

• Looked back at Biometric standards for ideas

  • ISO 30107-3 presentation attack detection - this is the typical system-architecture diagram that biometric folks are familiar with

Open

• Discussion about how to bring the diagram into the discussion - so that we can speak to the biometrics-oriented audiences

• Dan: there might be arrows missing if we overlay the IDPV flows and consider deepfakes - e.g. during enrollment, capture of the reference may go via the signal processor box into the data storage box.

• Jay: where do injection attacks happen in this diagram?

  • Dan: for image processing, injections happen at #2 before templatization

• Mike notes that this diagram was originally created for the fingerprint

• Stephanie has an updated diagram that addresses deepfakes

Open

• Joey: points out that the new world is both the biometric sampling and also now reference document modifications

  • Synthetically generated ID Documents should be in scope

• Sean: a big part of the threat is the ability to scale the attacks - important considerations

• Andrew: points out that we have to address physical document fakes during the transition towards digital ID documents (10-15 years)

• This content area fits into “attack vectors”

  • Volunteers: Max, Dan

  • Need to add in new boxes for how images are brought into the reference database - e.g. morph image attacks

•  

What to do with Heather’s content topics?

Jay

• Discussion about how the “hacker motivations” topic areas should be handled

  • An option to merge into other content areas - e.g. who is a target/is vulnerable

  • Look for opportunities to include “motivations” into the other topic areas

A Book with content on cyber-crime motives: https://www.goodreads.com/book/show/38502106-industry-of-anonymity

• Discussion about “synthetic identity and simulated humans“

  • These are key topics for the overall topic - needs descriptive and definitional work

  • Need to figure out how to talk about this - example given of fake LinkedIn profiles - generated

  • Perhaps start with taxonomy and deepen the content around e.g. synthetic imagery

  • (Look for Treasury’s glossary)

• Must read: Chapter 1 - https://link.springer.com/book/10.1007/978-3-030-87664-7

  • Springer’s handbook on digital face manipulations

• https://www.socure.com/glossary/synthetic-identity-fraud

• https://www.paravision.ai/resource/authentic-identity-challenges-and-opportunities-in-physical-and-digital-domains/

•  

Next week topic

Dawid: To present on forensics

Joey: to demonstrate live injection of deepfake into Zoom call

Places to follow

• 404 Media https://www.404media.co

• also their podcast