AI News Roundup — April 9, 2026
Anthropic quietly deploys its most powerful model through a secret defensive security program called Project Glasswing, Meta debuts Muse Spark under Alexandr Wang, and Intel surges after joining Elon Musk's Terafab AI chip project.
Anthropic Deploys Claude Mythos in Secret Project Glasswing With 40 Companies
Anthropic's Claude Mythos Preview is so powerful at finding and exploiting software vulnerabilities that it will not be released publicly. Instead, it is being deployed through Project Glasswing, a limited partnership program with over 40 companies including Microsoft, Amazon, Apple, Google, NVIDIA, CrowdStrike, and Palo Alto Networks, restricted to defensive security use only.
Meta Debuts Muse Spark, Its First Major AI Model Under Alexandr Wang
Meta launched Muse Spark, its first major model developed under chief AI officer Alexandr Wang, who joined from Scale AI nine months ago to lead Meta Superintelligence Labs. Meta announced its AI-related capital expenditures for 2026 will be between $115 billion and $135 billion, nearly double last year's spend.
Intel Surges After Joining Elon Musk's Terafab AI Chip Project
Intel joined Terafab, a massive semiconductor fabrication project led by Tesla, SpaceX, and xAI, aiming to produce AI chips at unprecedented scale. Intel shares surged on the news, marking a significant strategic pivot for the chipmaker as it seeks relevance in the AI hardware race.
Google Launches Gemma 4 With Frontier Performance on a Single H100 GPU
Google launched Gemma 4, a family of four open-source models that run entirely on a single 80GB Nvidia H100 GPU while delivering benchmark performance comparable to models 20 times their size. The Apache 2.0 licensed models represent a major advance in open-source AI efficiency.
AI Scribes Are Raising Healthcare Costs and No One Agrees on a Fix
Health insurers and hospitals now agree that AI documentation scribes are increasing healthcare costs, but diverge sharply on solutions. The debate highlights unintended economic consequences of AI adoption in clinical settings and raises questions about how to govern AI tools in healthcare.
Opinion: An AI Threat Looms and We Are Not Prepared
A widely-shared opinion piece warns that society, governments, and institutions remain dangerously unprepared for the pace of AI capability growth, particularly in cybersecurity and autonomous decision-making domains. The piece comes as Anthropic's own Mythos model is being withheld from public release due to its offensive cyber capabilities.