Meituan releases 1.6-trillion-parameter agent model trained on Chinese hardware
Meituan has released LongCat-2.5, a large-scale agent model trained entirely on Chinese hardware, with performance comparable to top US models on coding benchmarks.
U.S. Treasury Secretary Scott Bessent raised concerns about Nvidia's ability to prevent restricted AI GPUs from reaching China, calling for accountability within the company.
AT&T is scaling its use of autonomous agents to redesign business processes, with nearly 500 workflows already in production and plans to triple that number next year.
OpenAI is reorganizing its robotics team, integrating researchers into other departments to focus on general-purpose models that may reduce the need for specialized robotics AI.
Anthropic recommends dropping outdated prompting practices to optimize performance with the new Claude Opus 5.5 model.
Google's Gemini 4 Carbon has surpassed GPT-6.1 Sol in the MathArena benchmark, but at a significantly higher cost.
Anthropic's CEO admits significant uncertainty about the company's future as it prepares for its planned IPO, highlighting risks tied to compute investments.
Ransomware attacks have spiked, with AI enabling more sophisticated and frequent cyber threats, particularly in international operations.
Jensen Huang's departure from Nvidia has prompted varied responses across Asia, with some governments confirming chip supply commitments and others addressing ongoing investigations.
Character.ai and Ollie join a growing list of AI initiatives that have shut down or pivoted, highlighting the challenges in the sector.
Over 115,000 signatures have been gathered for a petition urging Nvidia to reinstate its former CEO, Jensen Huang, following his resignation amid controversy.
Nvidia's security team is under increased scrutiny as the company navigates leadership changes and external investigations.