AI Update, March 28, 2025: AI News and Views From the Past Week
Mar 28, 2025 - marketingprofs.com
The article provides a roundup of recent developments in the AI sector, highlighting significant advancements and strategic moves by major players. DeepSeek has launched an upgraded V3 model, enhancing its competitive stance against OpenAI by focusing on reasoning and coding capabilities. Tencent introduced its T1 reasoning model, aiming to improve AI speed and comprehension, while OpenAI integrated image generation into ChatGPT, streamlining visual content creation. Meta faced ethical scrutiny for using pirated books to train its AI, raising concerns about IP rights. Meanwhile, Elon Musk's xAI entered the visual AI space with a new image generation API, and Mistral's CEO predicted AI's potential to significantly boost global GDP.
The article also covers Google's release of its advanced Gemini 2.5 AI model, which enhances reasoning and multimodal capabilities, and its expansion of Gemini Gems to mobile users. Microsoft introduced new AI agents within Microsoft 365 Copilot, and Anthropic won an early round in a copyright lawsuit over AI training data. Deloitte opened a global AI simulation center in India, and a judge allowed The New York Times's copyright lawsuit against OpenAI to proceed. Additionally, OpenAI and Meta explored AI partnerships with India's Reliance, potentially expanding AI accessibility in the Indian market.
Key takeaways:
DeepSeek's upgraded V3 model intensifies competition with OpenAI, emphasizing lower operational costs and strong performance in reasoning and coding.
Tencent's T1 reasoning model aims to enhance AI capabilities in China, focusing on speed and long-text comprehension.
OpenAI integrates AI image generation into ChatGPT, streamlining visual content creation for users.
Meta's use of pirated books for AI training raises ethical concerns about intellectual property rights.
