AI News Roundup — April 13, 2026
Meta officially releases Muse Spark to widespread deployment and a 6.5% stock surge, while Anthropic announces its revenue has overtaken OpenAI at $30 billion ARR. Anthropic also reveals its Claude Mythos model is too dangerous to release publicly, and three Chinese AI firms are named as the primary actors behind large-scale model theft.
Meta Officially Releases Muse Spark — Stock Jumps 6.5%
Meta Platforms officially released Muse Spark, its first major AI model in over a year, scoring 52 on the Intelligence Index. The model delivers multimodal reasoning for health tasks and consumer applications, and is being deployed across WhatsApp, Instagram, Facebook, Messenger, and Meta AI glasses. Meta stock surged 6.5% on the announcement, outpacing the S&P 500 and Nasdaq on the day.
Anthropic Revenue Overtakes OpenAI at $30 Billion ARR
Anthropic's annualized revenue has reached $30 billion, tripling from $9 billion at the end of 2025 and surpassing OpenAI's $25 billion ARR. Nearly a third of American businesses are now paying for Anthropic's AI offerings, reflecting a significant enterprise shift toward Claude as a preferred AI platform.
Anthropic Says Claude Mythos Is Too Dangerous to Release Publicly
Anthropic disclosed that its Claude Mythos Preview model is so advanced at finding software vulnerabilities that the company fears it could become a hacker's most powerful tool if released broadly. The model has uncovered tens of thousands of critical software flaws across every major operating system and browser, and will remain restricted to vetted cybersecurity partners only.
OpenAI Releases GPT-5.3 Instant Mini as New Fallback Model
OpenAI rolled out GPT-5.3 Instant Mini inside ChatGPT as a new fallback model, offering more natural conversation, stronger writing, and better contextual awareness at lower cost. Pro and Plus plans received updated Codex usage options, and OpenAI simultaneously launched a new $100/month Pro tier with higher usage limits and priority access to advanced models.
OpenAI, Anthropic, and Google Name Chinese Firms Behind Model Theft
The three major AI labs officially identified DeepSeek, Moonshot AI, and MiniMax as the primary actors behind systematic model theft via adversarial distillation. Anthropic revealed these firms collectively generated over 16 million exchanges with Claude through approximately 24,000 fraudulent accounts. Intelligence is now being shared through the Frontier Model Forum to counter ongoing theft attempts.
2026 Responsible AI Symposium Convenes Industry, Government, and Academia
Leaders from technology companies, government, and academia gathered at the 2026 Responsible AI Symposium to address growing governance challenges. Discussions focused on liability frameworks, transparency requirements, and coordinating international standards before diverging national regulations fragment the global AI market.
Wave of State AI Laws Targets Deepfakes, Disclosure, and Autonomous Systems
A new wave of state AI laws took effect or advanced in legislatures this week, targeting AI-generated deepfakes, mandatory disclosure requirements, and autonomous decision-making systems in employment and healthcare. The measures set up a looming conflict with federal preemption efforts as the White House and state legislatures spar over jurisdiction in governing AI.