Overview
- Anthropic said it disrupted in September an operation that used an AI system to help automate elements of hacking against about 30 global targets, linked by the company to actors tied to China, drawing skepticism from security researchers and denials or no comment from Chinese officials.
- Peer‑reviewed research in Nature Communications found large language models shifted policy views as effectively as human‑written arguments in three experiments with 4,829 U.S. participants, raising concerns about scalable, personalized political influence and prompting calls for disclosure and platform detection tools.
- Enterprise use keeps climbing, with Stanford HAI reporting 78% of organizations now deploy AI; industrial cases cited include Schneider Electric projects that cut farm energy costs by up to 50% and a Nescafé plant in Mexico avoiding costly unplanned outages through analytics.
- Media and fact‑checking leaders warned of rising AI‑driven misinformation and error rates in automated answers, while consumer studies report roughly 60% prefer content created or overseen by identifiable humans and 71% would drop brands that simulate testimonials with AI.
- Experts urge role‑specific training, human‑in‑the‑loop procedures and proportionate regulation as companies test agentic commerce pilots with voice assistants and connected cars, and new surveys show uneven adoption such as Argentina’s 45.5% personal and 27.5% workplace use with notable demographic gaps.