Alibaba and China Telecom launch a massive AI data center in Guangdong powered by 10,000 homegrown Zhenwu chips, signaling China's aggressive push toward semiconductor self-sufficiency as U.S. export restrictions cut off access to advanced Nvidia hardware.
X is rolling out Grok-powered auto-translation globally and an AI photo editor on iOS with natural language editing capabilities, but consent concerns loom as it remains unclear whether safeguards will prevent sexualized edits of others' images.
Perplexity AI's annual recurring revenue surpasses $450 million as of March 2026, more than doubling in a single quarter, while its user base tops 100 million and its valuation soars to $21 billion — up from just $3 billion in mid-2024.
OpenAI launches a Child Safety Blueprint in partnership with child protection organizations, as AI-generated child sexual abuse cases surge over 8,000 in early 2025, outlining urgent steps to strengthen detection, legislation, and built-in AI safeguards against exploitation.
HeyGen's new Avatar V AI system outperforms all tested rivals in lip-sync accuracy and realism, using a Diffusion Transformer and novel Sparse Reference Attention to capture identity and dynamic behaviors, achieving win rates between 68.9% and 85.7% in human evaluations.
Anthropic launches Claude Managed Agents in public beta, a powerful suite of composable APIs enabling developers to build and deploy cloud-hosted AI agents up to 10x faster, with major companies like Notion, Asana, and Atlassian already shipping production-ready agents in days instead of months.
Google Finance is rolling out AI-powered market research, real-time earnings call insights, and advanced charting tools to over 100 countries, bringing live corporate earnings transcripts and local language support to global investors.
Windsurf launches SWE-1.6, a blazing-fast AI coding model hitting 950 tokens per second, now freely available for three months, featuring smarter task execution with reduced overthinking and faster parallel tool calls powered by Fireworks and Cerebras.
TriAttention, a new open-source method, compresses KV cache by 10.7x and doubles inference speed on long reasoning tasks while maintaining full attention accuracy, using trigonometric frequency-domain compression and integrating seamlessly with vLLM for easy local deployment on memory-constrained GPUs.
A new open-source toolkit called Gemma Multimodal Fine-Tuner now lets developers fine-tune Google's Gemma AI models on text, images, and audio directly on Apple Silicon Macs, requiring no NVIDIA hardware and supporting terabyte-scale cloud datasets while keeping data fully private on-device.