type: timeline_event
On March 17, 2026, Bloomberg reported that the Pentagon's Chief Digital and Artificial Intelligence Office (CDAO) was actively working to develop alternatives to Anthropic's Claude and transition classified AI workflows to competing platforms. Cameron Stanley, a senior CDAO official, confirmed that the office had begun the transition process and was evaluating multiple replacement options. Pentagon CTO Emil Michael acknowledged that the transition would take "more than a month" but insisted it was achievable within the six-month deadline President Trump had set for eliminating Anthropic systems from government networks.
OpenAI had already been deployed in certain Pentagon environments and was positioned as the nearest-term replacement, with models being tested in intelligence analysis and planning applications. Google's Gemini was identified as the next system in the pipeline, with integration work underway for defense and intelligence use cases. The speed with which the Pentagon was pursuing replacements signaled that the administration intended to make the blacklist permanent regardless of the outcome of Anthropic's legal challenge, treating the transition as a fait accompli.
However, the reporting also revealed significant technical challenges. Claude had been deeply integrated into classified workflows over the preceding two years, and replacement was not a simple matter of swapping one language model for another. Custom fine-tuning, security certifications, and workflow-specific integrations would all need to be rebuilt from scratch. Defense contractors who had built products on Claude's API — including Palantir's Maven system — would need to recertify their platforms with new underlying models, a process that insiders warned could take far longer than Michael's optimistic timeline suggested. Breaking Defense reported growing skepticism within the defense technology community that a clean transition could be completed without significant operational disruption.