Expanding Solutions to Address the Risks of Face Recognition Systems

Today we are publishing the first of a series of discussion papers regarding face recognition and biometrics. The ideas in this paper will be discussed in the ID4Africa live cast, Spotlight on Face Recognition. 

Too often, discussions about what to do about the risks of face recognition have lacked the kinds of specific protections for the use of face recognition that exist in abundance in other contexts, such as drug safety or chemical safety. These robust, effective and time-tested regulatory solutions have been omitted from the discussion of how to address face recognition risks. This has left a gap.

To make progress, it will be necessary to mature and expand the regulatory dialogue around face recognition systems and biometrics. Adapting practices from existing, useful safety models is an important step to bringing in workable, tested ideas for addressing products that pose risks. This paper describes these practices and discusses the most useful protective components. There is no need to reinvent the wheel — the current safety regulation models already in use can be adapted to work quite well for biometric technologies.

    • Chemical safety regulations manage dangerous substances with administrative, procedural, and other meaningful, robust controls with clear accountability. 
    • The model for bans changes in these models. Bans exist, but are not ad hoc or political, nor are they are common. There are meaningful procedures to be followed that lead to a ban. These procedures are based in science, fact, meaningful regulatory oversight, and multistakeholder work. This work is non-adversarial. 
    • These regulatory models are in use globally. 
    • Africa, Asia, EU, US, India, and almost all other jurisdictions all have meaningful regulatory safety frameworks 
    • WHO and other global multilateral institutions track these types of safety regulations closely.

In short, there is a lot to learn from well-established chemical safety regulations, and much that can apply to face recognition systems and improve outcomes.

Article: Face Recognition Systems: Expanding solutions by using time-tested safety models to address face recognition risks (PDF, 9 pages) 

Related Work: