Anthropic Breaks Ranks With Big Tech Lobby ITI Over AI Chip Export Controls

In a dramatic break with the traditional tech policy establishment, AI frontier lab Anthropic severed its membership with the Information Technology Industry Council (ITI) on September 8, 2026. The departure came immediately after the powerful tech trade group urged congressional leaders to strip three landmark semiconductor export-control measures from the annual National Defense Authorization Act (NDAA) package. By stepping away from a coalition that represents giants like Google, Nvidia, and OpenAI, Anthropic has firmly aligned itself with national security hawks demanding stricter controls on advanced computing power, directly challenging the industry consensus that pushes for broader commercial access abroad.

Background

For decades, the Information Technology Industry Council (Information Technology Industry Council - Wikipedia) has served as one of Silicon Valley's most influential voices on Capitol Hill, lobbying on tax, trade, intellectual property, and national security policies for the world's largest hardware and software providers. Historically, trade associations present a unified front to lawmakers, arguing that voluntary industry standards, minimal regulation, and open international markets drive American innovation.

In recent years, however, the intersection of advanced semiconductors and cutting-edge artificial intelligence has transformed commercial silicon into strategic geopolitical assets. The U.S. government has increasingly relied on export controls to curb the flow of frontier AI chips to foreign adversaries, primarily China, out of concern that advanced processors could accelerate autonomous weapons development and foreign intelligence capabilities.

While hardware manufacturers rely on global supply chains and foreign sales for revenue, frontier AI developers like Anthropic have argued that the computational power (or "compute") necessary to train and operate advanced artificial intelligence systems is a dual-use weapon requiring strict containment. The resulting friction came to a head over the Senate manager's amendment to the defense authorization bill, which incorporated three bipartisan export-security proposals.

What Happened

Anthropic officially walked away from ITI following a letter sent by the trade group to the House and Senate Armed Services Committees lobbying against three legislative provisions: the AI OVERWATCH Act, the Chip Security Act, and the MATCH Act, as first reported by Axios and AI Weekly.

These three bills were designed to close existing loopholes in current export control regimes:

• The AI OVERWATCH Act: Would mandate that the U.S. Department of Commerce notify Congress and introduce a mandatory review window before approving export licenses for sensitive dual-use chips destined for adversarial countries.
• The Chip Security Act: Would require chipmakers to build location-verification and tracking mechanisms directly into advanced AI chips to combat global smuggling networks.
• The MATCH Act: Targets critical semiconductor fabrication chokepoints by pushing international allies to restrict sales of advanced chipmaking equipment to China.

ITI’s push to strip these measures reflected the interests of legacy hardware firms that view stringent hardware tracking and license delays as burdensome impediments to business. Nvidia, an ITI member, has historically criticized such measures, arguing that firms should prioritize innovation over legislative interventions targeting chip smuggling.

Anthropic, by contrast, strongly backed the three acts. The company's leadership—including CEO Dario Amodei, who previously advocated for robust chip containment in a Wall Street Journal op-ed—believes that without statutory teeth, existing chip controls lack durability. Despite ITI having previously supported Anthropic during its legal clashes with the Pentagon regarding deployment limitations on its Claude models (Crypto Briefing), the startup opted to leave the lobby rather than remain tied to its export agenda.

Why It Matters

Anthropic’s exit shatters the long-held myth that Silicon Valley speaks with one voice on national security and AI governance.

Hardware vendors like Nvidia depend on shipping physical processors, while hyperscalers like Google provide extensive cloud infrastructure. Frontier model labs, however, are increasingly concerned about model distillation, algorithmic theft, and the proliferation of frontier models trained on smuggled compute. If foreign labs can acquire equivalent compute clusters, the competitive moat and safety guardrails established by domestic developers are largely compromised.

Furthermore, this move underscores Anthropic's willingness to operate as a rogue agent among its peers. The company had already parted ways with groups like OpenAI and Google on state-level safety standards; now, it is directly confronting their federal trade representation over hardware policy. For lawmakers, having a premier AI developer validate national security concerns provides bipartisan cover to dismiss corporate lobbying as protectionism rather than genuine technological insight.

What's Next

All eyes now turn to the final negotiations over the defense authorization package. ITI’s lobbying push may still find traction among defense lawmakers worried about bureaucratic slowdowns in chip licensing, but Anthropic’s public walkout has dramatically raised the political stakes. Stripping the AI OVERWATCH Act, Chip Security Act, and MATCH Act now risks appearing as capitulation to hardware lobbies against the explicit warnings of leading AI researchers.

In the medium term, Anthropic’s break highlights the growing likelihood of a permanent realignment in tech lobbying. As frontier AI models move closer to national security infrastructure, the divergence between software safety advocates, consumer tech platforms, and semiconductor manufacturers will only widen. Observers should watch whether other AI safety-focused organizations follow Anthropic's lead to form distinct policy coalitions, fundamentally reshaping how Washington regulates artificial intelligence.