tech governance round-up #2
This week,
South Korea is set to establish the "Korea AI Action Plan," which specifies the direction of AI policies to respond to the expansion of domestically focused AI ecosystems in the US and China. The action plan will specifically focus on building an AI ecosystem, including the specific supply of advanced graphics processing units (GPUs), plans to attract foreign talent, enhance AI utilization in the public sector for the national AX transformation, and establishment of AI factories. South Korea also established and launched its National AI Strategy Committee to centralize governance and accelerate innovation. The committee expanded the previous Presidential Committee on AI and set it up as a command centre with the authority to supervise strategies, coordinate across ministries, and evaluate AI projects. The committee now has 50 members, including a private sector representative and ministers from the finance, education, science, defence, industry, and data protection sectors.
In the United States, Michigan passed a new law to ban all forms of AI-generated pornography. Penalties include up to 1 year in jail and/or a fine of up to USD 3000, and where an incident is considered a felony, the penalty will include up to 3 years in prison and/or a fine of up to USD 5000. Ohio became the first state in the US to require public schools to adopt policies for the use of AI, particularly to create rules on how AI can be used responsibly.
Japan is set to formulate its first basic AI policy plan by Winter 2025. It is intended to both accelerate adoption and enable domestic leadership in AI. On September 1, the new AI Strategy Headquarters was launched, and is led by Prime Minister Shigeru Ishiba. The AI policy is aimed at addressing risks like misinformation, data security, misuse of AI, and enabling innovation as driver of economic growth.
China rolled out a new law requiring all AI-generated text, images, audio, video, and other virtual content to carry explicit and implicit labels. All explicit labels must be visible to users, whereas implicit identifiers like digital watermarks must be embedded in metadata. The regulation was drafted by the Cyberspace Administration of China along with three other ministries, and is part of the larger 2025 Qinglang campaign to tighten the oversight of AI, combat misinformation, and prevent fraud.
India’s Central Drugs Standard Control Organisation is streamlining regulations using AI in order to improve efficiency. This is aimed at reducing bureaucratic hurdles, setting fixed approval timelines, and shifting toward lighter regulations with stronger execution. With this, processing times for drug test licenses have been reduced to 45 days and export certifications are cleared in under 7 days.
The Western Cape Government approved the first provincial AI policy framework for public healthcare in South Africa. This policy sets ethical guardrails to balance innovation, patient safety, and equity within the health sector, and is driven by the key principles of beneficence, non-maleficence, human oversight, fairness, and transparency. The framework is intended to expand access, reduce waiting times, and protect the rights of patients. The policy suggests that AI has the potential to reduce administrative burdens, prevent medical errors, and bridge language gaps.