Three stories define the last weekend of June 2026 in AI: a blockbuster Information investigation reveals that well over half of all Grok traffic is pornographic content; an enterprise spending revolt threatens the growth model of Anthropic and OpenAI ahead of their planned IPOs; and tomorrow, Anthropic hosts its biggest AI-for-science event yet.

Grok's Adult Content Strategy: A Deliberate Choice

The Information published a deep investigation this weekend revealing that over 50% of all traffic flowing through Grok is adult content, pornographic image generation, explicit video creation, erotic roleplay chat, and large volumes of erotica requests. The report describes this as a deliberate strategic choice rather than an accidental outcome.

While OpenAI, Anthropic, and Google each prohibited explicit content generation across their consumer-facing products, xAI moved to fill the gap, investing engineering resources in expanding Grok's image and video generation capabilities specifically in directions its competitors declined to pursue. Users also discovered a pricing arbitrage: routing adult content requests through Grok's coding-focused models costs less than the primary consumer interface.

The legal exposure is significant. SpaceX set aside approximately $530 million in legal reserves to cover litigation costs from the adult content strategy. xAI engineers told The Information they found no reliable way to block child sexual abuse material generation without dismantling the explicit content business entirely. The European Commission has launched an investigation into X over sexualized deepfakes; criminal investigations are underway in France; and multiple US lawsuits are pending.

The traffic numbers tell the story: Grok web traffic fell 22% from January to May 2026, the steepest decline of any major AI platform. Over the same period, Claude traffic grew 369% and DeepSeek grew 44%.

The AI Spending Revolt Is Real

CNBC published a parallel investigation into a shift that has been accelerating for months: enterprise customers are switching from Anthropic and OpenAI to cheaper alternatives, primarily DeepSeek, because token costs have become unsustainable for production applications.

Flo Crivello, CEO of AI startup Lindy, told CNBC he switched his company off Claude entirely, moving 100% of its traffic to DeepSeek. "We did it, and you could see that cost curve go down, like, crash to the ground." The decision will save Lindy millions of dollars within months. "It's a matter of survival for the business. That's all it is."

Gil Luria, equity analyst at D.A. Davidson, told CNBC: "Current growth rates for Anthropic and OpenAI are the fastest they will ever be, which is mostly a matter of basic math. That is a good reason to go public now, as is the concern that some of their largest enterprise customers may start limiting their out-of-control token spend."

Microsoft CEO Satya Nadella wrote in June: "The last thing any of us want is a world where every company across every sector is ceding value to a few models that eat everything they see." Microsoft unveiled a suite of new low-cost models and emphasized that GitHub Copilot will route users to the most appropriate model for a task, not necessarily the most expensive one.

Both Anthropic (confidential S-1 filed June 1) and OpenAI (confidential S-1 filed June 8) are preparing IPOs at near-trillion dollar valuations built on current growth rates. The CNBC investigation spotlights the key risk: those growth rates may be at or near their peak.

Tomorrow: Anthropic's AI for Science Event

Monday June 30 at 10:00 AM PST, Anthropic hosts "The Briefing: AI for Science", the most substantive AI-for-science event the company has ever organized. John Jumper, Nobel Prize 2024 in Chemistry, AlphaFold co-creator, and nine-year Google DeepMind veteran who joined Anthropic, is expected to make his first public appearance since the move.

The event arrives backed by real research. In June, Anthropic published VirBench, a benchmark of 120 viral sequence retrieval queries. Claude Sonnet 4 scored 16.9% accuracy in standard conditions. After Anthropic built a deterministic retrieval tool (gget virus), accuracy jumped to 92.8%. The lesson: deterministic data infrastructure matters more than model capability alone.

Also Today

SK Hynix continues preparations for a $29.4 billion US ADR listing on Nasdaq, among the largest secondary offerings in history, to raise capital for expanding HBM memory chip production capacity for AI workloads. Microsoft expects total 2026 capex to reach $190 billion, with $25 billion of the increase attributed to surging memory and storage component prices.

The bottom line: The AI industry stands at a crossroads, on one side, a capability arms race among frontier labs with tightening regulation; on the other, mounting economic pressure from enterprises that refuse to pay premium prices for tasks that cheaper models handle adequately. The upcoming Anthropic and OpenAI IPOs will be tested not just on valuation, but on whether their meteoric growth rates are already behind them.