AI News Roundup — June 3, 2026
Trump signed a landmark AI executive order requiring voluntary 30-day government previews of advanced AI models, while SpaceX set its IPO price at $135/share targeting a record $75 billion raise. Microsoft Build Day 2 continued with Mustafa Suleiman's keynote on Microsoft AI's full stack, and WWDC 2026 looms just five days away with Apple's rumored AI overhaul.
Trump Signs AI Executive Order: Voluntary 30-Day Government Model Review
President Trump signed an 'AI Innovation and Security' executive order requiring AI companies to voluntarily submit advanced AI models — particularly those with significant cyber or national security capabilities — to the federal government for review up to 30 days before public release. The order is described as voluntary, not mandatory. It is a scaled-down version of the broader AI EO that Trump postponed last month after pressure from Musk and Zuckerberg. The order also directs federal agencies to accelerate AI adoption across government operations, framing the US approach as 'innovation-first' vs the EU's 'regulation-first' model.
CNN · NPR · New York Times
SpaceX Sets IPO Price at $135 Per Share — Targeting Record $75 Billion Raise
SpaceX set its IPO share price at $135, targeting the largest IPO raise in US history at $75 billion. The roadshow began June 3 ahead of a planned June 12 listing. SpaceX's S-1 disclosed a Q1 2026 net loss of $4.28 billion — up more than 700% year-over-year — as the company invests heavily in Starship and orbital AI compute infrastructure. Reuters warned 'beware the ghosts of IPOs past' as SpaceX, OpenAI, and Anthropic all prepare for public markets simultaneously at valuations that imply continued hypergrowth.
Microsoft Build Day 2: Mustafa Suleiman Keynote, Microsoft IQ, and Rayfin
Microsoft AI CEO Mustafa Suleiman delivered the Day 2 keynote, focusing on the full AI stack for enterprises. Key launches: Microsoft IQ (generally available today), a unified context layer combining Work IQ (organizational intelligence from emails, meetings, documents), Fabric IQ (structured business data), and Foundry IQ (retrieval planning across enterprise and web). Rayfin, a new enterprise-grade app backend, lets developers build AI-native business applications without managing infrastructure. GitHub Copilot's agentic development app launched for pull request workflows. Agent 365 and MDASH (multi-model agentic security) were also demoed.
Trump AI Order Context: Why Anthropic Welcomed It and OpenAI Didn't
The new executive order is essentially a Rorschach test for each AI lab's situation. Anthropic welcomed it — voluntarily submitting to government review aligns with its safety-first positioning and could reopen the path to federal contracts after the Pentagon blacklisting. OpenAI resisted, arguing it creates competitive disadvantage by exposing model capabilities pre-launch. The order's 'voluntary' framing means labs can effectively opt out, but the reputational and regulatory pressure to participate is significant.
WWDC 2026 Preview: Apple's Siri Overhaul and AI-First iOS Arrives June 8
Apple's WWDC 2026 keynote is five days away, and the previews are pointing to the most significant Siri and iOS redesign in years. Apple is reportedly rebuilding Siri from the ground up, powered by a custom Gemini model under a $1 billion/year licensing deal with Google. iPhone 18 is expected to ship with a fully conversational, multi-turn Siri alongside iOS 27. Analysts expect Apple to demonstrate on-device AI capabilities powered by Apple Silicon, competing directly with the Nvidia RTX Spark AI PC category that Microsoft just entered.
AI Film-Making Reaches Inflection Point — '$2,000 for What Would Have Cost Millions'
The Guardian profiled 'Dreams of Violets,' an AI-generated feature film that its director said would have cost millions in CGI to produce conventionally but was made for $2,000 using AI video generation tools. The piece captured a debate splitting the film industry: is this the democratization of cinema or AI slop that undermines creative workers? The same week, the WGA warned that its new AI protections in the studio deal do not cover AI-generated directors' cuts and pitch visuals.
Anthropic EU Cybersecurity Agency Gets Mythos Access — White House Remains Opposed Domestically
Anthropic granted the EU's cybersecurity agency ENISA access to Mythos Preview for defensive vulnerability research, marking the first time a foreign government security body has received access. The move signals Anthropic is willing to use international partnerships to advance Mythos deployment despite the domestic impasse — the White House remains opposed to broader US public access, fearing the model's dual-use capabilities.