WA police facial recognition trial scanned 130,000 faces
Australia's first live police facial recognition deployment is quietly scanning tens of thousands of people in Perth — and the backlash is already here. Also in this brief: Beijing anoints its first official lab for embodied-AI models, and Korea's KT commits billions to a 1GW data center buildout.
Western Australia Police's live facial recognition trial has scanned more than 130,000 people in Perth and Fremantle since June, in the first deployment of its kind by an Australian police force. Using NEC's Neoface m40 system, cameras mounted on police vans compare faces in real time against a police watchlist — the more complex "one-to-many" matching that privacy advocates have long warned turns every bystander into a potential database entry. The pilot has drawn rebukes from privacy experts and legal advocates who point to an opaque watchlist, unclear data-retention rules, and trial locations that some say disproportionately cover areas with high First Nations populations. WA Police call the results promising, and the force is being closely watched by law enforcement agencies across the country as the trial runs toward an official assessment. The stakes go beyond one state: this is Australia's live test of whether real-time mass facial recognition in public spaces is acceptable — and its outcome will shape policy in every other jurisdiction watching.
Beijing has approved China's first provincial-level key laboratory dedicated to embodied-intelligence large models, with robot startup Galaxy General leading it alongside Peking University and CATL. The lab, unveiled in Haidian on August 9, is the only officially recognized provincial-level research center in the "embodied large model" field — the frontier where AI models gain bodies and learn to act in the physical world. Galaxy General founder and CTO Wang He will serve as director, with an academic committee chaired by Chinese Academy of Engineering academician Gao Wen, and a nine-year roadmap running from whole-body end-to-end model breakthroughs to industrial deployment. The involvement of CATL, the world's largest battery maker, signals that Beijing sees embodied AI as an industrial pillar rather than a research curiosity — and the lab's creation is another marker of how aggressively China is formalizing state backing for humanoid robotics.
South Korea's KT plans to invest about 5 trillion won ($3.5 billion) over five years to secure 1GW of AI data center capacity, part of a broader pivot into the compute economy. The carrier's wider 18 trillion won five-year plan to become an "AX platform company" allocates 8 trillion won to network infrastructure, 4 trillion won to IT and security, 5 trillion won to data centers, and 1 trillion won to submarine cables. KT is not alone: SK Telecom is targeting up to 5GW of capacity through its SK Hyper unit, and LG Uplus is expanding its Paju campus, as Korea's telcos bet the AI infrastructure buildout can offset a stagnant telecom core. The familiar hurdles — power, land, and financing — apply, but the direction is unmistakable: carriers are repositioning themselves as the landlords of the AI era.
What to watch: WA Police's final trial assessment, due after the pilot's performance review, and whether Australia's privacy watchdog tightens rules on live facial recognition in public before other states follow suit.
Would you accept real-time police facial recognition in your city if it came with strict oversight? Tell us in the comments.
