Corporations implement guardrails on AI access amid soaring spending and security concerns

Major corporations are shifting from encouraging broad AI adoption to implementing stricter controls, including usage budgets and access restrictions, as AI investment escalates. Goldman Sachs CEO and chairman David Solomon noted that while expanding AI tools for its 45,000 employees, not all staff require access to the most advanced models daily, particularly outside of cybersecurity needs. Other firms are tightening policies after initial permissive periods. Cisco's chief product officer, Jeetu Patel, mentioned that after allowing coding teams unlimited tokens to build fluency, managers across the 32,000-person team are now receiving token budgets. JPMorgan, for instance, imposed a monthly $2,000 cap on some engineers' use of Anthropic's Claude Code, citing productivity and risk reduction. This oversight follows periods of massive usage, such as Visa doubling its monthly token use from 1 trillion in February to nearly 2 trillion in March. Furthermore, a 2025 KPMG survey indicated that 44% of over 1,000 US workers used AI without authorization. Rippling, a workforce-management startup, set monthly caps after projecting token spending could consume 40% of its R&D budget. A new Menlo Ventures study revealed enterprise AI deployment spending hit $177 billion by September, up significantly from $33 billion in all of 2025.