Enterprise readiness · Governance security

Anthropic Reports Analysts Followed Agent Preferences

Oct 27, 2026, first seen via OpenAI & Anthropic publications. Covered by 9 outlets.

Anthropic reported that professional analysts followed a randomly assigned AI coding agent preference 86% of the time in a randomized trial.

Read the original at anthropic.comOpens the publisher's site in a new tab

More in Enterprise readiness

Microsoft Offers Buyouts To 60,000 EmployeesOct 26 Parents Sue OpenAI Over ChatGPT Medication GuidanceNov 1 Anthropic Reaches Copyright Settlement Over Training BooksOct 27 OpenAI Will Honor Shared Training SignalsOct 30 Cognizant Commits 25000 Professionals To OpenAIOct 30