Anthropic CEO Dario Amodei has proposed creating an AI regulatory framework modeled on agencies such as the US Federal Aviation Administration (FAA) to address growing public safety and national security concerns. According to Business Standard, Amodei argues that while AI is advancing at extraordinary speed, governments and legislatures are moving much more slowly. He warns that if this trend continues for another year or two, the world could see what he calls 'Powerful AI', equivalent to 'a country of geniuses in a datacentre'. The proposed framework includes mandatory testing for advanced AI models above a compute threshold, assessing risks in cybersecurity, biosecurity, AI autonomy and automated R&D, and allowing governments to block or delay unsafe models. The framework would also require strong security measures, red-teaming and threat monitoring, with government agencies or accredited independent bodies conducting evaluations. In his latest essay, Amodei wrote that 'It is time to go beyond transparency to more serious and binding regulation of AI', comparing AI regulation to cars, airplanes, or drugs - powerful technologies essential to the modern economy but capable of killing large numbers of people if designed or operated poorly.
Under the new framework, frontier developers would be required to test models before release and publish summaries, safety frameworks, and system cards for powerful AI systems. Independent evaluators would review model tests and risk reports, with at least one qualified evaluator required to review company evaluations and publish findings on model risk reports. The framework calls for at least one qualified independent evaluator with governments and industry setting standards for evaluators. Developers would also maintain strong security programs for model weights and training systems, with companies describing their security programs publicly at a high level and sharing more details with designated government agencies when requested. Business Standard reports that Amodei has proposed AI regulations should match the rigor of Federal Aviation Administration regulations, with AI models required to undergo technical testing and auditing like airplanes before release. In his latest essay, Amodei emphasized that AI models should undergo mandatory testing by third parties to assess the risk they pose across several categories, including enabling cybersecurity threats and biological weapons, with the government having the power to block or deter deployment if AI is deemed to 'present unacceptable risks'.
Anthropic has launched a $200 million Economic Futures Research Fund to examine how AI will reshape employment and economic growth, while urging governments to prepare for potentially severe labour market disruptions. According to Anthropic, CEO Dario Amodei argued that AI could transform the workforce more rapidly and more extensively than previous technological revolutions. The company outlined three levels of economic disruption that could result from AI adoption, including scenarios where unemployment rises to 5%, 10%, or reaches levels beyond historical precedent - significantly higher than the current 4.3% US unemployment rate. For the most severe scenario, Anthropic suggested governments may eventually need long-term redistribution mechanisms, including basic income programmes, sovereign wealth funds and equity-sharing arrangements that allow the public to benefit directly from AI-generated wealth. As reported by Business Standard, Amodei proposed universal basic income mechanisms that could be financed through taxes on relevant companies or by raising capital gains tax. The latest unemployment rate reported last week was 4.3%, making the current scenario a baseline for potential AI-driven disruption.
Anthropic has identified four main risk areas that the framework addresses: biological risk, cyber risk, loss of control, and automated AI research. For biological risk, the company warns that unsafe systems could help attackers develop harmful viruses while similar AI tools can support drug discovery. For cyber risk, frontier models can find serious software flaws at large scale, raising concerns for hospitals, energy grids, and other key systems. The framework also highlights risks from systems acting outside developer control. Automated AI research could increase biological, cyber, and control risks if safeguards fail, with the company calling for better tools to detect, contain, or shut down unsafe systems. Business Standard reports that the company's proposals add that the government should be able to 'block or deter' the rollout of AI models that 'pose a significant risk of catastrophic harms'. Amodei warned that cybersecurity risks may only be the beginning, with biological threats and serious AI autonomy risks following as systems become more capable. In his latest essay, Amodei cited Mythos as a prime example of 'AI's incredible power, as well as its risks', noting that the company previously announced the more powerful AI model capable of spotting and exploiting vulnerabilities in critical software, before deciding to limit its release to select partners.
Anthropic CEO Dario Amodei has warned that powerful AI could become an unprecedented tool for authoritarian governments if appropriate safeguards are not established. According to Business Standard, he argued that AI-enabled surveillance could allow governments to analyse vast amounts of public information and infer highly personal details about citizens, capabilities that existing civil liberties frameworks were never designed to address. Similarly, future autonomous weapons systems could enable governments to exercise power with reduced human oversight. To safeguard democratic institutions, Amodei proposed clear accountability rules for fully autonomous weapons, a ban on domestic deployment of autonomous weapons systems, closing data broker and bulk data collection loopholes, and ensuring citizens have access to AI-based assistance. He also argued that AI should not be entrusted entirely to either governments or corporations, calling instead for checks and balances on both. Regarding geopolitical concerns, Amodei advocated for a coalition of democratic nations built around shared AI values, including free sharing of advanced chips and semiconductor manufacturing equipment, coordinated management of AI supply chains, and mutual defence against adversarial AI systems. The intervention comes as President Donald Trump outlined a hands-off approach to addressing cybersecurity threats raised by AI under an executive order that calls for giving the US government voluntary access to models, but stopped short of requiring developers to seek explicit approval.
The intervention places Anthropic among a growing number of AI companies calling for broader discussions on how the wealth generated by artificial intelligence should be distributed. Earlier this week, OpenAI said it wanted the gains from AI to be 'widely shared', while OpenAI CEO Sam Altman has explored proposals for public ownership stakes in AI firms through a national wealth fund. Speaking at the White House on Wednesday, US President Donald Trump said he would soon meet executives from major AI companies to discuss ways of ensuring the public benefits from the industry's growth. 'We're talking about giving back something to the public, and if we do that, the public will become very rich,' Trump stated. As part of its policy framework, Anthropic outlined measures including stronger data collection to measure AI-related job losses, incentives that encourage companies to retain workers, and broader economic support if automation causes sustained employment declines. Business Standard reports that the announcement comes on the heels of Anthropic rival OpenAI outlining goals that included ensuring gains from the technology are 'widely shared'. Keeping in line with these concerns, Amodei highlighted that Anthropic is backing proposals on frontier AI model testing and job displacement, while supporting transparency legislation in California, New York and Illinois.
