ByteDance implements strict ban on external AI tools for employees

ByteDance is reportedly implementing a comprehensive restriction on the use of external artificial intelligence tools across the company, effective this month. This new policy prohibits the use of personal accounts for services like Claude and Codex within internal systems, specifically targeting L4-level code repositories and Feishu documents to mitigate data leakage and security risks. This heightened control is focused on protecting core assets, as L4 repositories house the fundamental architecture of ByteDance’s business lines. Previously, starting in June 2025, the company had already begun phasing out third-party coding aids such as Cursor and Windsurf, mandating the use of its proprietary assistant, Trae. Vice President of Engineering Hong Dingkun noted the severe consequences of leaking core content at L3 and above. This move aligns with a broader trend among major tech firms prioritizing data compliance. The decision is reinforced by founder Zhang Yiming’s assertion at a Seed team meeting that the company would not rely on AI distillation technology to improve its models, even if it meant temporarily falling behind competitors.