The debate: Are facial recognition cameras in Sainsbury's a step too far? (www.bbc.co.uk)

Sainsbury’s deployment of facial-recognition cameras in stores has reignited a debate about retail surveillance: proponents argue the technology can deter and identify repeat offenders and reduce shrinkage, while critics warn it’s a heavy-handed, poorly regulated intrusion with real harms. The discussion brings together voices from law enforcement, lived experience, civil liberties campaigning and academia—former Met DCI David Spencer sees crime-reduction potential; Keeley Knowles, a former shoplifter now working in drug outreach, stresses that many thefts stem from addiction and social causes that cameras won’t fix; Matthew Feeney of Big Brother Watch calls the measure disproportionate; and AI ethicist Kate Devlin highlights algorithmic bias and the lack of legal safeguards. Technically and operationally, the concerns center on biased models and false positives producing wrongful matches, opaque vendor systems, data retention and sharing practices, and the absence of clear regulatory standards for accuracy, auditing and accountability. For the AI/ML community this raises practical issues: how to build and benchmark fairer identification systems, how to quantify and mitigate disparate impact across demographics, and how to design transparent audit trails. The consensus among critics is that any deployment should be paired with strict legal limits, independent audits and consideration of non-technological interventions to address root causes of theft.