US warns China’s “industrial-scale” AI copying—while Brazil probes Google’s news AI use

On April 23, 2026, Michael Kratsios, the White House director of the Office of Science and Technology Policy, said in a memo that China-based foreign entities have run “industrial-scale” campaigns to steal or replicate frontier AI models from US companies. The claim is framed as an intelligence and technology-transfer problem rather than ordinary competition, and it ties directly to US concerns about model theft, training data access, and IP leakage. In parallel, the White House accused Chinese companies of copying US AI technology, reinforcing a pattern of escalating public attribution. Separate from the US-China thread, Brazil’s CADE (Conselho Administrativo de Defesa Econômica) unanimously decided to open an investigation into Google over its use of news content by AI tools. Strategically, the US allegations aim to harden the frontier AI perimeter by treating model extraction as a national-security issue, not merely a corporate IP dispute. This shifts the power dynamic toward tighter export controls, procurement screening, and potential restrictions on cloud, data, and compute access for high-risk actors. China benefits if it can accelerate capability gains through cheaper replication, but it also faces reputational and regulatory costs as US officials build a case for broader defensive measures. The Brazil decision adds a different but complementary layer: it signals that AI deployment in media and information markets will face competition-law scrutiny, potentially reshaping how model providers monetize content and partner with publishers. Together, these moves suggest a world where AI supply chains are increasingly governed by security and antitrust constraints, not just engineering performance. Market and economic implications are likely to concentrate in AI infrastructure, software, and information-services ecosystems. US-listed AI and cloud beneficiaries could see near-term sentiment support if investors expect stronger enforcement and procurement preferences, while companies exposed to IP or data-sharing risks may face higher compliance costs and potential litigation. The CADE probe can pressure Google’s advertising and licensing economics in Brazil by raising uncertainty around licensing terms, revenue sharing, and data access for AI-driven news products. In commodities and FX, the direct linkage is indirect, but heightened technology tensions typically lift risk premia for semiconductors, networking equipment, and cybersecurity spend, with spillovers into USD risk sentiment. Watch for volatility in AI-adjacent equities and for changes in credit spreads tied to legal and regulatory overhang. Next, the key watchpoints are whether the US escalates from public attribution to concrete instruments—such as sanctions, export-control tightening, or restrictions on specific AI training and deployment pathways. For markets, the trigger is any follow-on action naming companies, detailing evidence, or expanding compliance requirements for frontier model developers and cloud providers. In Brazil, the investigation’s scope—whether it targets licensing practices, scraping/ingestion methods, or algorithmic ranking—will determine how quickly publishers and platforms renegotiate terms. A practical timeline is the next CADE procedural steps and any US interagency updates following Kratsios’s memo, which could arrive within weeks and intensify if additional incidents are reported. De-escalation would look like clearer evidence thresholds, narrower enforcement, or negotiated licensing frameworks that reduce uncertainty for AI news products.

Geopolitical Implications

• 01

  AI governance is merging with national security, increasing the likelihood of restrictive policy tools.

• 02

  US-China competition is shifting toward regulatory and procurement leverage.

• 03

  Brazil’s antitrust stance may reshape AI content ingestion and licensing globally.

Key Signals

• —Named enforcement actions following the White House memo
• —Evidence thresholds and company-specific allegations
• —CADE’s scope and remedies for Google’s AI news practices
• —Publisher-platform licensing announcements in Brazil