AI Mornings with Andreas Vig
Your daily AI news briefing in under 10 minutes. New models, product launches, research breakthroughs, and industry shifts, explained clearly, no hype.
AI Mornings with Andreas Vig
Anthropic's IPO Filing & Microsoft's Break from OpenAI
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
Hey! Welcome to AI Mornings with Andreas Vig. It's June 3, 2026. Anthropic has officially filed to go public. The company confidentially submitted its draft S1 registration statement to the SEC yesterday, putting it ahead of OpenAI in the race to the public markets. This comes just days after Anthropic raised $65 billion at a $965 billion valuation, making it the most valuable AI startup in the world. The IPO will be one of the largest tech offerings in history, and it'll give the public its first real look at Anthropic's financials. What's particularly interesting is that Anthropic chose to move now, while OpenAI is still reportedly months away from its own filing. The CloudMaker has been gaining serious ground in enterprise and coding applications, and this move suggests they're confident in their position. Also making big moves today, Microsoft just launched MAI Code OneFlash, its first in-house coding model built specifically for GitHub Copilot. This is a significant shift because, until now, GitHub Copilot has relied heavily on OpenAI models. The new model outperforms Claude Haiku 4.5 across multiple benchmarks, including a 16-point lead on Suite Bench Pro, but what's really notable is the efficiency. Microsoft says the model solves harder problems with up to 60% fewer tokens. That translates to lower costs, faster responses, and a smoother experience for developers. The model is rolling out now to GitHub co-pilot users in Visual Studio Code, and it's trained on clean licensed data with the GitHub Copilot production harness built directly into its training. Microsoft is clearly building its own AI stack, and this reduces their dependence on OpenAI. Meanwhile, Anthropic is expanding Project Glasswing, its cybersecurity initiative that uses Claude Mythos to find vulnerabilities in critical infrastructure. They're adding about 150 new organizations across more than 15 countries, covering power, water, healthcare, communications, and hardware. The initial partners have already found over 10,000 high or critical severity security flaws. What's striking here is Anthropic's warning. They expect other AI companies to have mythos class cyber capabilities within 6 to 12 months, and those models could be released without proper safeguards. The company is essentially saying that powerful offensive cyber tools are coming whether we're ready or not, and defenders need to adapt now. On the enterprise front, OpenAI is pushing deeper into white-collar work with six new Codex plugins targeting specific professional roles. There are tools now for data analytics, creative production sales, product design, equity investing, and investment banking. Codex has more than 5 million weekly active users now, and knowledge workers represent about 20% of that base, growing three times faster than developers. OpenAI also launched a sites feature that lets Codex output work as hosted interactive websites with partnerships including Wix, Replit, and Figma. The enterprise push is clearly accelerating. But here's a reality check on enterprise AI adoption. Uber just capped employee AI spending at $1,500 per month per tool after burning through its entire annual AI budget in just four months. The company had encouraged staff to use AI as much as possible and even ranked usage on internal leaderboards. Now they're pulling back. Uber's COO recently said it's very hard to draw a line between AI usage and actual productivity gains. That's a sentiment we're hearing more often, and it raises a real question for the industry. Where's the ROI? In policy news, President Trump signed an executive order on AI oversight yesterday, but it's significantly watered down from earlier drafts. The final version asks AI companies to voluntarily submit new models for government review 30 days before public release. An earlier draft had called for 90 days, and industry figures, including former White House AI czar David Sachs, pushed back hard. The order explicitly states it doesn't create any mandatory licensing or preclearance requirements. Basically, it's a polite request with no teeth, which tells you how much leverage the AI industry currently has in Washington. Here's a study that caught my attention. Researchers at Stanford Law School found that law professors overwhelmingly prefer AI-generated answers to student questions over answers written by their peers. In nearly 3,000 blind comparisons, AI won 75% of head-to-head matchups. These weren't simple factual questions either. They were contract law questions requiring nuanced reasoning and judgment. Perhaps most striking, professors flagged AI responses as pedagogically harmful only 3.5% of the time, compared to 12% for peer-written answers. This is one of the first rigorous studies showing AI can meet the professional standard in a field that requires real judgment, not just knowledge retrieval. On the consumer side, Google is rolling out fake call detection on Android to protect against AI voice impersonation scams. The feature uses a kind of digital handshake between devices. When a contact calls you and you're both using Google's phone app, their device sends a silent confirmation signal to verify the call is actually coming from them. If a scammer tries to impersonate your mom using AI voice cloning, that signal will be missing and you'll get a warning to hang up. It's rolling out globally this month, starting with pixel devices. This is Google directly addressing one of the more dangerous applications of generative AI. And finally, Martin Scorsese has partnered with AI image startup Black Forest Labs as an advisor. The 83-year-old director is using the technology for storyboarding, saying it helps him communicate his vision faster than his 70 years of hand-drawn boards. Black Forest Labs was founded by the team behind Stable Diffusion and is valued at $3.25 billion. It's another sign that Hollywood's resistance to AI is softening even among its most legendary figures. That's all for today. Thanks for listening, and I'll see you tomorrow.