OpenAI, Anthropic face pressure for AI safety amid $7T scaling push

OpenAI: OpenAI is a leading AI research and deployment company developing advanced artificial intelligence systems. It is preparing to go public while engaging in discussions about AI safety and self-regulation alongside competitors. The company faces internal conflicts over balancing rapid model improvement with safety concerns, as highlighted by former researcher Daniel Kokotajlo. Anthropic: Anthropic is an AI company focused on building reliable and interpretable artificial intelligence systems. It is preparing to go public at a high valuation amid industry-wide debates on safety oversight and model transparency. Like peers, it navigates pressure to advance capabilities while addressing risks from complex AI deployments. Goldman Sachs: Goldman Sachs is a major global investment bank providing financial services and economic analysis. It has been referenced in coverage of massive investments in AI scaling and the resulting competitive pressures on frontier labs. Its analysis highlights the economic forces prioritizing speed in AI development over safety measures. Daniel Kokotajlo: Daniel Kokotajlo is the executive director of the AI Futures Project and a former OpenAI researcher who departed the company in 2024 over safety concerns. He argues that AI cannot be safely scaled and advocates avoiding further model improvements to force a slowdown. Kokotajlo has shared his research through talks with current industry employees on internal conflicts around AI direction. Akshay Krishnaswamy: Akshay Krishnaswamy is the chief architect at Palantir, specializing in data analytics and AI software. He has stated that human willpower can enable safe AI scaling, including through the use of open models. His perspective contributes to discussions on reconciling safety with rapid AI advancement. • A 2026 study reported that intense competition can push AI firms to allocate more resources to speed and less to safety. • Anthropic CEO Dario Amodei called for slowing frontier development so safety measures can keep pace. • OpenAI, Anthropic, and Google DeepMind have discussed coordinating on AI-safety standards and evaluations. • The International AI Safety Report identifies liability, insurance, procurement, and regulatory incentives as tools that could improve private-sector safety investment. • Industry leaders disagree on whether commercial incentives and legal liability are already sufficient to prevent dangerous releases. • The headline frames trillions of dollars in potential AI value as a structural incentive for market domination, rather than evidence that every company neglects safety.