AI Industry Daily News
A roundup of the AI industry's day, centered on Codex Windows support, grok-build-0.1, Claude Opus 4.8, Command A+, and Rosalind Biodefense.
Today's highlights
Key topics and reactions
Trump Floats Plan to Give Americans Stakes in AI Companies
Speaking to reporters aboard Air Force One on June 5, 2026, Trump said there is "a concept where stakes are given to the American people, who would essentially become partners in the companies." He said the administration would examine the idea and planned to meet with major AI companies at the White House next week.
CNBC reported that OpenAI CEO Sam Altman and the White House have discussed government stakes for over a year, in line with the "Public Wealth Fund" concept OpenAI proposed in an April policy paper. Reported candidates include OpenAI, Anthropic and xAI, though Trump named no specific firms.
The proposal comes as AI companies prepare IPOs while public skepticism remains high, with polls showing many Americans believe AI's harms outweigh its benefits. The administration has already taken a 10% stake in Intel, signaling a departure from traditional free-market orthodoxy.
White House AI Policy Chief Sriram Krishnan to Step Down at End of June
Krishnan announced on social media that he will depart the White House this month, without detailing his reasons. After a break, he said he plans to launch a new institution to tackle "the big challenges around American AI."
Appointed by then-President-elect Trump in December 2024, Krishnan worked closely with AI and crypto czar David Sacks on the American AI Action Plan, an executive order on a National AI Policy Framework, and AI summits with allies in France, India, the UK and the Middle East.
A former general partner at a16z with product experience at Microsoft, Twitter, Yahoo, Facebook and Snap, Krishnan shaped industry-friendly policy emphasizing the dominance of the "American AI stack."
Anthropic Calls for a Global Pause on Frontier AI Development
Anthropic's call for a temporary halt to frontier development on safety grounds has reignited debate across the industry. A company co-founder attended the Vatican's release of its first official AI doctrine, a document under Pope Leo emphasizing a "human-first" principle.
Separately, evidence of an upcoming Claude Sonnet 4.8 has accumulated, with two of three leaked strings reportedly confirmed, raising expectations for the next model.
The moves coincide with broader soul-searching over AI's social impact, paralleling the Vatican's doctrine and rising public skepticism toward the technology.
Anthropic Secures $35B Debt Package to Buy and Lease Google TPUs
The package is structured as debt across three tranches rather than equity. A special purpose vehicle borrows the funds to buy Google TPUs and lease them to Anthropic, with Broadcom backstopping the senior portion to strengthen the deal's credit.
Initially floated at around $36 billion, the deal closed at $35 billion in early June 2026. The financing aims to accelerate data-center-scale TPU leasing as Anthropic races OpenAI, Google and Microsoft in the AI arms race.
Anthropic has previously raised over $8 billion from Amazon and multiple rounds from Google, using AWS and Google Cloud as primary infrastructure. Reports place its valuation between $380 billion and $965 billion amid IPO speculation.
Category highlights
Bot Traffic Overtakes Humans Online
Cloudflare CEO Matthew Prince said on June 3, 2026 that bot traffic surpassed human traffic on the internet for the first time, driven by surging agentic AI. Cloudflare data showed bots at roughly 57.4–57.5% of HTML HTTP requests versus 42.5–42.6% for humans. Prince said the reversal arrived more than 18 months ahead of his SXSW prediction, as agents visit dozens to thousands of pages per query.
Replit Hits $9B Valuation
Replit, the "vibe coding" platform, raised a $400 million Series D at a $9 billion valuation, tripling its $3 billion mark from July 2025 in six months. Founder and CEO Amjad Masad's net worth reached an estimated $2 billion. The Georgian-led round drew a16z, Coatue, QIA and celebrities including Shaquille O'Neal and Jared Leto. Replit unveiled "Agent 4," a digital-canvas tool for non-engineers, targeting $1 billion ARR by end of 2026.
Developers Shift From Prompts to Loops
Boris Cherny, Head of Claude Code at Anthropic, said he no longer prompts AI directly but runs a "loop" that lets Claude decide what to do next, arguing engineers' work is moving toward writing loops. Anthropic now writes most of its code with the tool; Cherny said he processed 259 fully AI-generated PRs in a month running 5–10 parallel instances. New features like /loop and Dynamic Workflows support the shift.
Verizon Says AI Lifted Customer Satisfaction
Verizon CEO Dan Schulman told Bloomberg Tech that a three-month experiment replacing some human operators with AI customer-service agents raised customer satisfaction by 1,280 basis points. AI handled routine tasks like password resets and billing, with complex cases handled by AI-human teams. The effort builds on Verizon's Google Cloud partnership using Gemini, which previously answered 95% of inquiries and lifted sales 40%.
Public Backlash Against AI Grows in US
The WSJ reported that negative sentiment toward AI is rising faster than the industry itself, citing data-center protests and polls. A Quinnipiac survey of 1,397 people in March 2026 found 55% believe AI does more harm than good in daily life (up 11 points), 70% expect job losses, and 65% oppose local data centers, citing power costs, water use and noise. Only 21% trust AI-generated information almost always.
Key trends
Vatican Issues First Official AI Doctrine
The Vatican released its first official doctrine on AI, with a document under Pope Leo emphasizing a "human-first" principle. An Anthropic co-founder attended the event, underscoring growing engagement between AI labs and institutions raising ethical questions about AI's societal impact.
CVPR 2026 Wraps Up
CVPR 2026 reached its final day with Thomas Serre's keynote "Scaling Laws vs. Neural Laws," Best Paper awards at the PAMI-TC meeting, and poster and oral sessions. Discussions noted US stagnation in video models, the rise of Chinese models like Seedance and Kling, and a shift toward omni models. An xAI/Vercel/Cursor-backed autonomous medical hackathon is set for June 13 in San Francisco.
Spotify Unveils AI Audio Tool "Studio"
Spotify announced an AI audio tool called Studio that generates personalized AI voice briefings from email, calendar and notes, with planned integration with ElevenLabs. Separately, PixVerse launched its "PixVerse Originals S1" series, producing the sci-fi dark comedy "Mars Landings" entirely with its tools.