News
Anthropic hands enterprises a "supervisory key": Claude can clearly see the AI spending of each employee. While the doubts about "whether AI will decide who to leave for the boss" are still lingering, Anthropic gave corporate customers a clearer management and control tool on October 29.
2 min read
Source: Telegram AI频道
Anthropic hands enterprises a "supervision key": Claude can clearly see the AI spending of each employee. While doubts about whether AI will decide who to leave for the boss are still lingering, Anthropic gave corporate customers a clearer management and control tool on October 29. The newly launched Claude smart report is open to Enterprise packages. Managers can open it with one click in the organization settings and generate charts to see at a glance what the team has done with AI, how much it has produced, and how much it has spent. Give the manager the "Oversight Key" to select the team, the longest 28-day interval, and the Claude products included in the statistics (conversation, Claude Code, Cowork or integrated interface, the newly released Duet AI is not supported yet), and the system will read the conversation, classify it into code, technical documents and other workflows and summarize the costs. Administrators can customize queries about usage and costs and save them for reuse, and the report will also recommend "which new rules should be added to the code repository." Click on any workflow in the chart to filter session details, costs, dates, and output types; even "conversations with no output content" and "chat marked as private" cannot escape the inefficiency mark. The first 10 copies per month are free, additional charges apply. Draw a red line: Do not evaluate performance, do not check privacy. This key will obviously touch the nerves of employees. There have been many reports of companies using AI to lay off employees. A recent survey showed that a quarter of managers in the United States "often or always" use AI to make such decisions. In July, former Meta employees accused the company of using insufficiently tested AI to evaluate production capacity and track word consumption to score layoffs. Meta denied it. Anthropic had already expected it. It clearly stated that the report is only used to find out the popularity of tools and planning investment. "The original design is not and should not be used to evaluate personal performance or personnel decisions." Sensitive dimensions such as race, gender, and health are prohibited. However, management and control capabilities and trust gaps coexist, making it difficult for employees to trust technology and managers with this set of tools. via AI News (author: AI Base)