Amazon's AGI lab head David Luan is departing the company to build something new, as the tech giant's AI division undergoes a major shake-up with leadership now shifting to senior vice president Peter DeSantis amid intensifying regulatory scrutiny over AI acquisitions.
Guide Labs open sources Steerling-8B, a groundbreaking 8-billion-parameter LLM that makes every generated token fully traceable to its training data, achieving 90% of standard model capability while giving developers unprecedented control over sensitive outputs — backed by a $9 million seed round from Y Combinator.
Google, Anthropic, and India are reshaping the AI landscape as Gemini 3.1 Pro shatters reasoning benchmarks, Claude Sonnet 4.6 brings million-token context windows to free users, and India commits $200 billion toward AI infrastructure in a massive push to become a global tech powerhouse.
A new autonomous AI agent fleet called SWE-AF is outperforming industry giants like Claude Code and Codex o3 on software engineering benchmarks, scoring 95/100 while costing as little as $6 per run using budget-friendly models.
Google Labs launches ProducerAI, a powerful generative AI music platform powered by Gemini and DeepMind models that lets anyone create full songs using natural language prompts, with Grammy-winning artists already on board and free global access available now at producer.ai.
Anthropic's new research reveals that AI models naturally develop human-like personas during pretraining — not through deliberate design — raising urgent concerns that developers must prioritize alignment from the very start of model development to protect users from psychological harm.
Moonshot AI launches Kimi K2.5, a powerful open-weight multimodal AI capable of vision processing and running up to 100 parallel sub-agents simultaneously, with benchmark scores rivaling GPT-5 and Gemini, and is now publicly available via API, chat, and Hugging Face.
IBM shares plummet 13.2% to $223.35, pushing year-to-date losses past 24%, after Anthropic's Claude Code tool threatens IBM's lucrative COBOL modernization business by automating complex legacy code analysis that previously took human analysts months to complete.
OpenAI warns Congress that Chinese AI rival DeepSeek is secretly stealing outputs from American AI models using obfuscated methods and third-party routers to train its R1 chatbot, raising urgent national security alarms in Washington amid growing concerns over chip export controls and Nvidia's reported technical support to DeepSeek.
DeepSeek unveils a powerful new sparse attention system in DeepSeek-V3.2 that uses a trained 'lightning indexer' to intelligently select the most relevant tokens, directly tackling the memory bottleneck slowing down long-context AI inference without sacrificing accuracy.