Anthropic announced that the automatic mode of Claude Code will be turned on by default. AI giant Anthropic recently announced that starting from August 14, it will turn on the "Auto Mode" (Auto Mode) of its programming tool Claude Code by default for Pro, Max and Team accounts.

Anthropic announced that it will turn on the Claude Code automatic mode by default. AI giant Anthropic recently announced that starting from August 14, it will turn on the "Auto Mode" (Auto Mode) of its programming tool Claude Code by default for Pro, Max and Team accounts. This means that programming with Claude Code will require less manual intervention. Anthropic first launched a test version of automatic mode in March this year, aiming to balance development speed and control. According to the company's official statement, in automatic mode, Claude Code will no longer prompt users for manual approval at every step of the operation. Instead, it will automatically advance the subsequent process unless the system determines that the operation is "irreversible, destructive, or targeted at the external environment." Regarding security considerations, the test data disclosed by Anthropic showed unexpected results: the automatic mode is more secure than manual review in actual applications. In a study of 1,053 paid testers, automatic mode successfully blocked 89% of harmful actions, while manual review only blocked 13.6%. Research analysis points out that this is mainly due to the fact that manual review is prone to habitual fatigue - when users use Claude Code on a daily basis, the approval rate for permission prompts is as high as 97%. Boris Cherny, the head of Claude Code, revealed on the social platform At the same time, Anthropic also added that the company has been continuously introducing new security protection measures, including prompt injection screening and custom hard rejection rules, to prevent security risks such as data leakage. Learn more: via cnBeta.COM - Chinese industry information station (author: Source: cnBeta.COM)