AI News Roundup: Mythos Spooks Wall Street, Google Drops Gemma 4, VC Funding Hits $300B

Bessent and Powell Summon Bank CEOs Over Anthropic’s Mythos AI

Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell convened an emergency meeting with the CEOs of America’s largest banks on Thursday to discuss cybersecurity risks posed by Anthropic’s Claude Mythos Preview. Executives from Citigroup, Bank of America, Wells Fargo, Morgan Stanley, and Goldman Sachs attended the session at Treasury headquarters in Washington. JPMorgan’s Jamie Dimon was the only major banking CEO absent.

The meeting comes days after Anthropic revealed that Mythos had discovered thousands of previously unknown zero-day vulnerabilities across every major operating system, web browser, and a range of critical infrastructure software — including a 27-year-old bug in OpenBSD. Anthropic launched Project Glasswing to deploy the model defensively with partners including Amazon, Apple, CrowdStrike, Microsoft, and the Linux Foundation, but officials want banks to prepare for a world where AI models of this caliber could fall into the wrong hands.

Google Launches Gemma 4 — Its Most Powerful Open Models Yet

Google DeepMind released Gemma 4, a family of four open-source models purpose-built for advanced reasoning and agentic workflows. The lineup includes E2B (2B parameters), E4B (4B parameters), a 26B mixture-of-experts variant, and a 31B dense model — spanning everything from smartphones and Raspberry Pi devices to cloud data centers. In a major licensing shift, all Gemma 4 models ship under a fully permissive Apache 2.0 license.

The models support multimodal inputs (text, image, and audio on edge variants), context windows up to 256K tokens, and over 140 languages. Benchmarks show the 31B dense model outperforming models 20 times its size, marking Google’s most aggressive challenge to Meta’s LLaMA in the open model arena. With over 400 million cumulative Gemma downloads, Google is betting that developer adoption will feed its broader Gemini ecosystem.

OpenAI, Anthropic, and Google Unite Against Chinese Model Copying

In a rare show of coordination, the three leading U.S. AI labs are working together through the Frontier Model Forum to combat adversarial distillation — a technique where competitors systematically query frontier models to train cheaper alternatives. Anthropic alone documented 16 million suspicious exchanges from three Chinese firms: DeepSeek, Moonshot AI, and MiniMax, routed through roughly 24,000 fraudulently created accounts.

The collaboration focuses on sharing detection methods and blocking terms-of-service violations at scale. The effort arrives as U.S.–China AI tensions escalate and the White House’s National Policy Framework for AI pushes federal preemption of state-level AI regulation to create a unified competitive posture.

Q1 2026 Venture Funding Shatters Records at $300 Billion

Venture capitalists invested a staggering $300 billion into 6,000 startups globally in Q1 2026 — up over 150% year-over-year and an all-time record. AI companies captured $242 billion, or roughly 80% of total funding. Four mega-rounds dominated: OpenAI closed $122 billion, Anthropic raised $30 billion, xAI secured $20 billion, and Waymo pulled in $16 billion. Together, those four deals represented 65% of all global venture capital.

U.S.-based companies claimed 83% of global VC, up from 71% a year earlier. Foundational AI companies — firms building core models rather than applications — raised $178 billion in the quarter alone, doubling the $88.9 billion raised across all of 2025. The concentration has reignited bubble concerns, but backers argue the infrastructure buildout justifies the scale.

Meta Commits Another $21 Billion to CoreWeave

Meta signed an additional $21 billion infrastructure deal with CoreWeave, on top of a prior $14.2 billion arrangement. The new agreement runs from 2027 to 2032, with CoreWeave’s data centers filled with hundreds of thousands of NVIDIA GPUs. Combined with its direct spending, Meta is now on track for up to $135 billion in AI infrastructure investment in 2026.

California Pushes Back on Federal AI Oversight

Governor Gavin Newsom signed an executive order asserting California’s right to independently assess AI companies’ risk designations, directly challenging the Trump administration’s approach. The move was triggered after the Department of Defense labeled Anthropic a supply-chain risk last month. Newsom’s order simultaneously encourages state employees to accelerate AI adoption while placing guardrails on its use — including a new AI-directed tool to help Californians navigate government programs and benefits.