Meta restricts employees' use of Claude Code and Codex over model distillation: Report

Meta has reportedly introduced restrictions on the use of Anthropic's Claude Code and OpenAI's Codex by engineers in its Applied AI division, amid concerns that interactions with rival AI systems could unintentionally lead to the replication of proprietary model capabilities.According to The Information, engineers in the Applied AI team now require approval before using the two AI coding assistants. The reported move comes as Meta works to reduce the risk that its own internal AI systems could interact with competing models in ways that may result in unintentional extraction of proprietary knowledge.Focus on model distillationThe concern centres on model distillation, a technique in which a smaller AI model is trained to reproduce the behaviour of a larger and more capable model. The approach is widely used to build models that are less expensive to run while retaining much of the performance of larger systems.However, companies developing advanced AI models have increasingly expressed concerns that extensive interactions with proprietary models could allow competitors to recreate important capabilities without direct access to the original systems. Such unauthorised replication has emerged as a legal and ethical concern in the AI industry.The reported restrictions mark one of the earliest publicly known instances of a major AI company limiting employee access to competing AI coding tools specifically over concerns related to model distillation.Meta has not publicly confirmed the reported restrictions. It also remains unclear whether the policy applies only to the Applied AI division or extends to other engineering teams within the company.The reported policy has also drawn attention to broader enterprise AI concerns. Although Anthropic and OpenAI offer enterprise versions of their AI products, questions remain over whether such offerings fully eliminate risks related to sensitive internal data or intellectual property being exposed during AI usage.Meta is not the only technology company to impose limits on external AI tools. Apple, Samsung and Amazon have previously restricted employees' use of ChatGPT or corporate GitHub Copilot over concerns that confidential code or trade secrets could be exposed through external large language models.