AI Expansion Raises Compute and Ethical Risks

AI is rapidly diversifying across sectors — from conservation and creative media to healthcare, cybersecurity and business — producing practical benefits alongside new risks. At Davos and in industry announcements, major corporate compute and capital deals have accelerated, including OpenAI’s disclosure of a 6 GW AMD GPU deployment and a warrant to buy up to 10% of AMD following a large Nvidia agreement, highlighting a tightly interlinked, potentially fragile compute–capital ecosystem. Commercial rollouts — such as generative-image/video tools like OpenAI’s Sora, Anthropic’s Claude 3.5, Snowflake AI agents, Google Cloud agent payments, and Apache Iceberg v3 — are moving AI from assistive tools toward more autonomous agents and broader enterprise use. That surge coexists with financial strains — including reports of large losses at OpenAI and warnings of a funding-driven correction that could lead to startup failures by 2026 — and with growing social concerns about privacy, data extraction, governance, transparency and vendor accountability; attackers likewise are using AI to automate sophisticated cyberattacks even as defenders adopt AI. Policymakers and leaders are urged to embed human judgment and oversight (for example, CMS’s WISeR pilot automating Medicare prior-authorization triage beginning January 2026), invest in “AI-proof” human skills, and upgrade governance and infrastructure to manage ethical, workforce and safety risks.