Category: TRENDS IN AI – Oct 20 2026

Following through on CEO Dario Amodei’s recent commitment, Anthropic is granting “employee-like” access to Model Evaluation and Threat Research (METR), a California nonprofit that evaluates the safety rails and risk factors of frontier AI models.

Engineers at the University of California Los Angeles have created an AI that uses light to analyze videos and spot deepfakes – videos that look real but were made by AI – correctly 98 percent of the time, Science magazine reported.

In a 12 October blog post, Anthropic raised the specter of a rogue AI assembling a botnet – a network of interconnected smart devices – that could expand at lightning speed, resist attempts to corral it, and hack into data centers to fuel its continued growth.

AI’s ongoing buildout will increase global growth at 3 percent this year but, at the same time, will create upheaval jobs, taxation, and education, Kristalina Georgieva, managing director of the International Monetary Fund (IMF) said on 12 October at the annual joint meeting of her agency and the World Bank.