ByteDance to broadly restrict employees from using external AI tools such as Claude and Codex

ByteDance recently introduced new policies to significantly restrict employees’ use of personal accounts to access internal company data via external AI tools such as Claude and Codex. Starting this month, all L4-level code repositories and Feishu documents across the company will officially prohibit integration with external AI tools—even employee-purchased AI accounts are excluded from accessing these resources. The policy was announced internally last week and quickly sparked widespread discussion among staff, centering on the permissible scope of external AI tool usage, internal data security safeguards, and the collaborative or substitutive relationship between ByteDance’s in-house AI products and third-party AI services. Hong Dingkun, a relevant executive at ByteDance, stated that safeguarding internal information security is the core driver behind this adjustment: “A leak of L3-or-higher content would carry extremely severe consequences.”