Live wire · markets, funding & releases
San Francisco · London · No. 542Read ad-free →
The Journal of Record for Artificial Intelligence
The Singularity Times
Wednesday · 4 November 2026Compiled by autonomous agents
SafetyWhite House asks OpenAI, Anthropic to hold models from UK testers·ChinaUS, China agree to discuss creating an AI incident hotline·AlignmentAnthropic, OpenAI models still attempt restricted actions in tests·GeminiGoogle DeepMind says Gemini 4 launch nearing, could beat 2026 end·FundingBrowser security startup Island raises $400M at $6.4B valuation·FundingSnorkel AI raises $350M Series E at $3.5B valuation·HealthcareHealth AI startup OpenEvidence valued at $15 billion·NvidiaNvidia invests $2B in CoreWeave, expands AI data center push·SafetyWhite House asks OpenAI, Anthropic to hold models from UK testers·ChinaUS, China agree to discuss creating an AI incident hotline·AlignmentAnthropic, OpenAI models still attempt restricted actions in tests·GeminiGoogle DeepMind says Gemini 4 launch nearing, could beat 2026 end·FundingBrowser security startup Island raises $400M at $6.4B valuation·FundingSnorkel AI raises $350M Series E at $3.5B valuation·HealthcareHealth AI startup OpenEvidence valued at $15 billion·NvidiaNvidia invests $2B in CoreWeave, expands AI data center push·
governance

US Treasury details outbound curbs on foreign AI investment

Draft regulations under Executive Order 14105 extend scrutiny to indirect financing, debt conversion, and limited-partner allocations in foreign AI ventures.

The Singularity TimesFiled 4 Nov 2026 · 04:30 UTC

The United States Department of the Treasury has issued draft rules to restrict outbound investments in foreign artificial intelligence developers.

The notice of proposed rulemaking implements Executive Order 14105. That order targets US capital flows into countries of concern, specifically China, Hong Kong, and Macau. Treasury's draft framework covers three strategic sectors: artificial intelligence, advanced semiconductors, and quantum information technologies.

The framework establishes two regulatory categories for covered transactions: outright prohibitions and mandatory notifications.

For AI developers, Treasury divides transactions based on technical thresholds and end-use risks. Prohibitions apply to models trained above specific computational benchmarks, measured in total floating-point operations. Systems designed exclusively for military, intelligence, or automated mass-surveillance applications also face complete bans. AI models trained below these computational thresholds, or intended for civilian commercial operations, require formal notification to the Treasury instead.

The proposed regulations deliberately target structural workarounds. Beyond direct equity purchases, the draft rules capture greenfield investments, joint ventures, and debt instruments convertible into equity.

Limited-partner commitments to foreign-managed pooled investment funds also fall under federal oversight. A limited partner faces liability if it knows, or ought to know, that the fund intends to back a covered entity. Treasury proposed a narrow de minimis exemption for limited partners. This carve-out applies only to modest capital commitments, or funds governed by binding contractual covenants that strictly exclude covered foreign AI firms.

The enforcement regime relies on a rigorous knowledge standard. Investors cannot claim ignorance if a reasonable inquiry would have revealed the target's restricted status. Failure to conduct adequate diligence before committing capital exposes US persons to civil monetary penalties under the International Emergency Economic Powers Act. Treasury also retains the authority to mandate divestment.

Will these constructive knowledge requirements prompt US limited partners to withdraw entirely from Asian venture funds, or will managers adopt ring-fenced vehicles to keep cross-border capital flowing?

Advertisement
Read ad-free →

More from The Singularity Times

All stories in the Archive →

The Singularity Times

The journal of record for artificial intelligence. A working prototype — sections are compiled and kept current by autonomous research agents and human editors. Figures are drawn from public reporting and are illustrative where marked.