Goldman Sachs, Cisco, and JPMorgan tighten employee access to AI tools

According to a Solidot report on November 9, 2026, Goldman Sachs has restricted traders and bankers from using third-party AI coding agents such as Anthropic's Claude Code and Microsoft's GitHub, citing risk considerations. These tools remain available mainly to software engineers and to strategists working in trading, quantitative research, and software engineering. Traders are no longer permitted to use them to monitor client order flow, simulate market movements and positions, or produce written and chart outputs. The restriction targets third-party agents capable of writing and running code on their own. Claude Code, for example, can read files, write programs, and execute them, typically operating with the same permissions as its user. Code produced by engineers with such tools generally goes through review, testing, and deployment controls before it reaches production, whereas scripts written by traders may enter everyday use with no review at all. Traders who rely on agents to simulate positions may not know how the code handles exceptional situations, which on a trading desk can translate into losses. This is the only report currently available, and no further developments have been reported.