MIT Technology Review is reporting that new research suggests artificial intelligence models can develop their own biases from experience, stereotyping job applicants more than humans. It said large language models already absorb human biases from training data, and the development of agentic AI risks exacerbating these issues.
MIT Technology Review is also reporting a rising risk of weather data sabotage. It said the growth of prediction markets and the shift to data-driven AI weather forecasting create temptations to manipulate data, potentially leading to systemic problems.
MIT Technology Review further revealed a new wave of high-tech luxury car theft, involving vehicles stolen during transport, which has been under the radar for two years.
Separately, The Wall Street Journal is reporting that SpaceX is negotiating to provide billions of dollars in AI compute capacity to the Pentagon. The Financial Times said Trump Media is pitching a $100,000 monthly fee to trading firms and banks for early access to Donald J Trump's social media posts. NPR reported that Immigration and Customs Enforcement shared Medicaid data it was not authorized to have with Palantir. The Washington Post said Washington is accelerating AI weapons development, opening the door to armed robots.
Full Article: The Download: AI hiring biases, and weather data sabotage