Agentic AI in 2026: Adoption and UC Berkeley Risk Framework

Summary

Agentic AI is moving from pilots to mainstream deployment. Gartner predicts 15% of day-to-day work decisions will be made autonomously via agentic AI by 2028, and 33% of enterprise software will include agentic capabilities (up from <1% in 2024). Anthropic research shows Claude Code users auto-approve agent actions over 40% of the time, with autonomous session length nearly doubling in three months. UC Berkeley's February 2026 framework addresses agentic-specific risks: unintended goal pursui