CANBERRA, Australia — Jeff Bleich, a former U.S. ambassador to Australia now serving as Anthropic’s international special envoy, is pressing Australia to play a larger role in shaping safety rules for advanced artificial-intelligence systems. Bleich discussed the company’s approach to AI governance at an AI masterclass hosted by the Australian Strategic Policy Institute’s Sydney Dialogues. His central argument was that the next phase of AI development will advance only as quickly as institutions can establish public confidence — in other words, at the “speed of trust.” The appeal comes amid increasingly stark warnings from researchers inside Anthropic and across the frontier-AI industry. Anthropic researchers have recently said that rapidly improving systems could, in extreme circumstances, escape human control. One senior alignment researcher estimated a 10% chance of human extinction within the next decade, while another said the industry lacks a credible plan for containing the most serious risks. Anthropic has emphasized that those estimates reflect the researchers’ personal views rather than official company forecasts. Chief Executive Dario Amodei has separately called for a slower and more internationally coordinated approach to developing frontier AI. The debate carries implications for Australia as it weighs how to regulate AI, protect democratic institutions and manage the technology’s effects on workers and communities. Cooperation between governments, technology companies and independent researchers could help establish common safety standards, but relying on corporate assurances alone would leave the public with little oversight of systems that may shape public services, workplaces and information ecosystems. Bleich’s role reflects Anthropic’s effort to build relationships with governments beyond the United States. His background as Washington’s ambassador to Australia gives him a direct connection to Australian policymakers as international governments confront how much authority to give companies developing increasingly powerful AI systems. The push for trust also highlights a political tension at the heart of the AI debate: Governments are being asked to support innovation while ensuring that companies do not set the rules for technologies whose risks may extend far beyond the firms that develop them.