The word “Ethical” was not mentioned once in this https://www.ncsc.gov.uk/collection/biometrics

While respecting stated security focus, failure to consider wider risks will continue to divorce functional success from harm.

Harm also occurs with entirely intentional and secure operation

This statement about scope is made. It decouples the guide from questions of bias in reference data sets and matching in practice. An essential risk to assess under core provisions of the new #EUAIReg if #Biometric #AI, but only if it impacts EU residents 2/8

This is first and foremost about functional fitness for identification, verification and authentication purpose - Who are you? Are you on the permitted access list?

Not so much authorisation - Limits on what and where you access once in. 3/8

#DataProtection and discrimination are mentioned. First in the "Costs" section, then under "Legal and policy constraints"

Note reference to PII not Personal Data. Link to a related thread below https://twitter.com/TrialByTruth/status/1369165003975180291?s=20

Here's another mention when exploring the use case of biometrics to unlock phones and apps. This refers to remote processing (comparing data to a database of biometric data held elsewhere)

This is one of the discussions about discrimination, through a functional lens #AIEthics #AIAudit

In statistical and systems limitation sense as justification for second authentication factor. Also a nod to historical issues with low enrolment for fingerprints

Then a further discussion about purchase considerations, like costs of enrolment and how they can be controlled with reuse of pre-existing data sets. That factor swinging it in favour of #FacialRecognition. Along with this statement about accuracy

All against backdrop of on-going lack of fit for purpose regulation and law in this space, plus law enforcement carve outs that can muddy waters for rules that do exist. Not bad, by many yardsticks, just devoid of wider legal, regulatory and ethical context, even via links

A thread as provocation to consider the impact of siloed thinking when managing risk. Over focus on declared intent and functional operation

Entirely and immediately relevant to #VaccinePassports Risk of use creep, data sharing, and #HumanRights erosion https://twitter.com/TrialByTruth/status/1378001383023054849?s=20

How (if you read the whole thing), would recommendations address pragmatic concerns raised in @OpenRightsGroup response to government #VaccinePassport consultation?

If work proposed by NCSC is not the time or place to cover it all, when is? https://www.openrightsgroup.org/blog/first-look-at-vaccine-passports/

Reverting to broader #AiEthics context, we are layering new functionality at scale onto weak foundations without reliable guardrails. Creating key system dependencies, with insufficient due diligence, exception management, and right to redress https://infospectives.wordpress.com/2019/06/06/in-ai-we-will-blindly-trust/

You can find out more about work to close this gap being done by @ForHumanity_Org via http://ForHumanity.center  where I also volunteer. My focus is refining #AiAudit and practical ways we might tackle #AiEthics risk /END