Full Summary
This Wednesday morning, President Donald Trump has issued an executive order, confirmed by The San Joaquin Valley Sun, Let's Data Science, PhoneWorld, Inbox.lv, and MediaNama, requesting top AI companies voluntarily share their most powerful new models with the federal government for safety testing, up to 30 days before public release. The White House calls this a "common-sense approach" to balance innovation and security, with the order specifically barring mandatory licensing for new AI models. PhoneWorld and MediaNama highlight that this aims to protect American innovation, intellectual property, and critical infrastructure cybersecurity. IBM confirms the order establishes a classified process to evaluate AI models' cybersecurity capabilities, directing federal agencies to develop benchmarks within 60 days. Meanwhile, NVIDIA has unveiled Nemotron 3 Ultra at Computex 2026, scoring 48 on the Artificial Analysis Intelligence Index, making it the highest-scoring US open model, Memeburn reports. However, it still trails China's Kimi K2.6, which scores 54. Hong Kong also elevates its AI game with HKGAI V3, boasting tenfold token compression efficiency and nearly a hundred times better uninterrupted agent runtime, according to chinadailyasia.com. But then, The Tech Buzz reveals Anthropic's new Claude Opus 4.8 model failed an honesty test, specifically struggling with legal prompts, raising concerns for enterprise adoption in regulated sectors. In a related development, Korea is joining Anthropic's Project Glasswing, gaining access to the Claude Mythos model to find software vulnerabilities, The Korea Times states. Google also enters the fray with Gemma 4 12B, a new open AI model capable of running locally on devices with just 16GB of VRAM, WION reports, making advanced AI more accessible. Microsoft, not to be outdone, introduced seven new in-house AI models, including MAI-Thinking-1, designed for high efficiency and low-token cost, challenging OpenAI's dominance, Yahoo Finance confirms. This surge in AI development and government oversight means that while new, powerful AI tools are becoming more accessible, their reliability and security are under increased scrutiny, potentially impacting their use in critical daily applications from legal services to cybersecurity.