South Asia Investor Review

Rogue AI agents are no longer just the stuff of science fiction. When an OpenAI rogue agent launched a cyber attack against AI startup Hugging Face this month, the company fought fire with fire, using an open-source Chinese AI model to defend against it.  During a test in a sandbox, an OpenAI autonomous agent managed to sneak out and attack a website operated by HuggingFace, an AI startup based in New York City. The OpenAI agent was not connected to the Internet but it found a way to access it and escaped its sandbox.  It took two days for Hugging Face to put an end to the attack, with help from GLM 5.2, an open weight system created by Chinese company Z.ai. All of this happened in the middle of a debate in the US about banning the use of Chinese AI models.  This incident gives one more reason to companies, including US-based firms, which are rapidly adopting  Chinese open-weight  AI models to save money. For example, Airbnb relies on Alibaba’s Qwen model, ...

On July 16 in Shanghai, 29 countries, including China, Pakistan and Russia, signed the founding agreement of WAICO, World AI Cooperation Organization.  Every BRICS founding member is in, except India. This agreement follows the launch of the US-led Pax Silica, a 24-member coalition, including  India , which is designed to counter China's AI efforts. The stated goal of both these competing groups is to provide global governance, including building guardrails and setting standards, for artificial intelligence (AI) technologies.  US-China AI Competition The announcement of WAICO coincided with the launch of Kimi K3 by  Chinese  startup Moonshot AI.  Kimi K3 is the largest open-weight AI model ever built, its full weights free for the world to download from July 27.  Cheaper open-source AI models like Kimi K3 pose a serious threat to U.S. proprietary models. By offering 80-90% of the capability at a fraction of the cost or for free, they are squeezing the ...

Pakistan's electricity demand has soared  21% in just two years . Rapid electrification is positively impacting all sectors of Pakistan's economy. thanks to growing deployment of distributed solar, estimated at 38 GW as of June, 2025. In 2025, 44% of solar deployment was residential, followed by industry (26%), agriculture (21%) and commercial users (9%). It is stimulating demand for a variety of products ranging from air conditioners and refrigerators to washing machines and electric scooters/motorbikes. As a result, Pakistan’s Large-Scale Manufacturing (LSM) sector grew by 6.5% in the fiscal year, rebounding from a –0.69% contraction the previous year.  The fiscal year saw a 61.66% surge in automobile production.  Impact of Solar Energy Revolution in Pakistan Dave Jones of   Ember , a global energy think tank, says Pakistan's experience with distributed solar could become the blueprint for dozens of developing countries.  If Pakistan is the first large-sc...

Policy-makers need data to formulate good policies. Good data produced by government agencies can be expected to lead to good policies and desirable outcomes. But data collection and statistical analyses require adequate methodologies and resources. Unfortunately, Pakistan's data quality gets a "C" grade by international agencies like the International Monetary Fund (IMF). Clearly the country faces significant data quality challenges.  These challenges range from estimation of the size and scope of the informal economy and electricity demand/consumption to education and nutrition.  Here are some examples of where the Pakistan Bureau of Statistics (PBS) data differs sharply from what is being reported by non-government groups:  1. Gross Domestic Product: A large chunk of Pakistan's economy is not documented . The PBS seems to be failing in making even the most rudimentary estimates of it. A 2024 joint study of the International Labor Organization and the Small and Medi...