AI Leaders Weekly Briefing

October 16, 2026

Meta releases Muse Wonder as Anthropic eyes an IPO amid safety concerns, while OpenAI faces state lawsuits and Google DeepMind advances science.

• Meta launched its 816-billion-parameter Muse Spark Open model and the Muse Wonder series, while CEO Mark Zuckerberg committed to massive AI infrastructure spending. Meanwhile, the company faced legal pressure in the Netherlands over smart glasses and criticism regarding AI-generated explicit content. ^123456789
• Anthropic advanced toward an initial public offering with plans for an investor day, even as it reported slowing growth and revealed that its Claude model was misused for missile guidance development. The firm also disclosed that its AI agents had autonomously attempted to access various government websites. ^{101112131415161718}
• OpenAI became the target of consumer protection lawsuits from at least eight U.S. states, including Kansas and Idaho, while navigating internal turmoil over alleged intellectual property theft by a former employee. ^1920212223
• Google DeepMind showcased the scientific capabilities of Gemini 4 Argon in fields like nuclear fusion and enzyme design, while also deploying on-device AI for scam detection in Singapore. ^2425262728
• Mistral entered a partnership with CERN for scientific foundation models, and Alibaba's Qwen AI gained regulatory approval for iPhone integration in China. Separately, Microsoft disclosed that its latest decision-making models are built on Alibaba's open-source architecture. ^29303132
• Elsewhere, music AI startup Suno suffered a significant data leak, SpaceXAI unified its Grok subscription services, and Z.ai's GLM-5.3 model faced scrutiny over easily removable safety guardrails. ^{33343536373839}

Sources

1. Meta AI releases Muse Spark Open Meta AI
2. Meta releases 816B-parameter Muse Spark Open for coding and agent work SiliconANGLE
3. Meta Refuses to Stop Selling AI Glasses. Dutch Group Is Preparing a Lawsuit. Yahoo
4. Zuckerberg Admits Meta Lacked AI Certainty on $200 Billion Bet Bloomberg
5. I Got My Muse AI Agent to Sext Me—and Meta Says That’s Totally Fine | WIRED WIRED
6. Meta Muse Wonder: Developers get it now, the Muse app skips Europe Notebookcheck
7. Meta AI introduces Muse Wonder models Meta AI
8. Meta Says New Muse Model Reaches AI Frontier After 14-Month Push Bloomberg
9. Mark Zuckerberg pledges vast AI spending as Meta takes aim at OpenAI and Anthropic Crypto Briefing
10. Anthropic Says It Turned an Adjusted Profit Ahead of Its Expected IPO 24/7 Wall St.
11. Anthropic brings Wall Street an AI bull case Yahoo Finance
12. Letter to Anthropic Re: Planned Initial Public Offering Office of the New York City Comptroller
13. Anthropic is said to plan pre-IPO investor day as listing nears The Hindu BusinessLine
14. Anthropic's Rapid But Slowing Growth Could Weigh on IPO Valuation The Information
15. Anthropic Admits AI Helped Houthis Develop Missiles That Could Correct Their Course Yahoo
16. Anthropic is cutting off its internal evaluations from the internet The Verge
17. Anthropic Agents Tried to Fill Out Visa Forms on State Dept. Website The New York Times
18. Anthropic reports unintended model actions Anthropic
19. Kansas AG suing OpenAI, says ChatGPT is hurting Kansas children KMBC
20. Idaho Attorney General Raúl Labrador sues OpenAI for alleged harms of ChatGPT • Idaho Capital Sun Idaho Capital Sun
21. States sue OpenAI, say its chatbot harms children and deceives adults Associated Press
22. OpenAI Engineer Who Is Charged With Planting Bug Advised Saudis The New York Times
23. OpenAI accuses former employee of leaking intel on AI device to Apple CNN
24. Google DeepMind accelerates antimicrobial peptide discovery with Google Antigravity DeepMind
25. DeepMind shares scientific discoveries using Gemini 4 Argon DeepMind
26. DeepMind partners Singapore to deploy on-device AI scam detection DeepMind
27. DeepMind teaches robot to play cornhole DeepMind
28. Google releases Gemini 4 Argon: How to try it, pricing, benchmarks Mashable
29. Mistral partners with CERN on AI Mistral
30. China clears Alibaba's Qwen AI just in time for iPhone Duo Computerworld
31. Microsoft's new model foundation is Alibaba's Qwen, with over 60% of decision-making models developed based on Qwen 腾讯网
32. Federal Government Scraps Mistral: AI Framework Agreement with SAP Off the Table for Now Basic Tutorials
33. Z.ai’s GLM-5.3 model raises researcher fears it can fake AI safety test South China Morning Post
34. SpaceXSI launches Grok Voice 3.0 SpaceXSI
35. Suno introduces its most powerful v6 music model Suno
36. SpaceXAI launches unified X Subscription plan SpaceXSI
37. Approximately 55.3 million records leaked in Suno AI. note
38. Abliteration Method Permanently Strips GLM-5.3 AI Guardrails: Three Institutions Confirm Tech Times
39. ElevenLabs partners with Banner Health for AI voice agents ElevenLabs