Asia tightens AI rules
Asia is moving quickly on AI governance: Taiwan’s Artificial Intelligence Basic Act is already in force (since January 2026), and Vietnam has rolled out a new AI law that mixes EU‑style risk thinking with stronger state oversight. At the same time the White House’s AI blueprint faces congressional hurdles over federal vs state authority, creating a fragmented regulatory landscape companies will need to navigate. (Law.asia) (TechPolicy.Press) (foxbaltimore.com)
Taiwan’s Legislative Yuan passed the Artificial Intelligence Basic Act on December 23, 2025, after nearly two years of debate over balancing innovation and rights. (bakermckenzie.com) The Act assigns the National Science and Technology Council (NSTC) as the central competent authority, requires the Executive Yuan to form a National AI Strategy Special Committee chaired by the premier, and charges the Ministry of Digital Affairs with building an internationally aligned risk‑classification framework. (bakermckenzie.com) The statute sets a phased implementation schedule that includes publishing minors/human‑rights/gender impact assessments within three months, completing government AI‑use reviews within six months, and tasking agencies to adopt AI use rules or internal controls within 12 months. (bakermckenzie.com) Vietnam’s National Assembly adopted Law No. 134/2025/QH15 on December 10, 2025, and the law is set to take effect on March 1, 2026, replacing most AI provisions in the 2025 Digital Technology Industry law. (tilleke.com) The Vietnamese law regulates by role (developers, providers, deployers, users), imposes a three‑tier risk classification (high/medium/low) that providers must apply before deployment, and requires high‑risk systems to maintain technical documentation, human‑oversight mechanisms and operational logs while MoST prepares an official high‑risk list based on five criteria. (mst.gov.vn) Vietnam also obliges providers to label AI‑generated audio, images and video with machine‑readable markers and to clearly identify content that replicates real people or simulates real events. (mst.gov.vn) The White House published a National Policy Framework for AI on March 20, 2026, urging Congress to preempt state AI laws it deems “undue burdens,” declining to create a new federal regulator, and calling for measures including a ratepayer‑protection pledge and streamlined permitting for AI infrastructure. (whitehouse.gov) Multiple outlets report the administration’s blueprint faces an uphill climb on Capitol Hill and is unlikely to win bipartisan support, with lawmakers suggesting any federal AI legislation could be tied to a separate kids’ online safety package. (politico.com)
