Z.AI's AI programming product ZCode has announced a new round of user compensation measures regarding the recent code data upload controversy, while also disclosing the latest progress on product rectification.

According to information released by ZCode, the compensation measures took effect at 10:30 AM on September 28.

For paying users and those who return as paying users within one month, ZCode will grant four weekly quota reset cards and four 5-hour quota reset cards, with the relevant benefits valid for one month.

In addition, from September 28 to October 7, ZCode will offer all users a daily giveaway of "100 million Tokens x 100,000 units."

On the relevant page, ZCode stated, "Trust cannot be reset with a single click. We will earn it back slowly, one improvement at a time."

Alongside the compensation plan, ZCode once again disclosed the rectification status of the earlier code data upload issue.

According to official information, the repository snapshot upload pipeline has now been removed, the cloud data involved has been deleted, and this has been confirmed through third-party verification.

ZCode also stated that without users actively initiating relevant operations, "nothing will leave the user's computer, and code remains local."

According to earlier reports, on September 18 Beijing time, multiple users of the AI programming tool ZCode, developed by AI technology company Z.AI, reported that the tool would "silently upload" users' programming and development data to third-party cloud servers without user permission.

The uploaded data included the entire historical record of the programming and development process, which may contain keys, internal business information, and other highly sensitive research and development assets and private information for developers.

Related users also reported that this data upload process could not be actively canceled by users in the client, and although the uploaded data was encrypted, users could not unpack it locally or through the client.

On the day of the reports, September 18, Z.AI apologized and attributed the issue to the "code repository indexing" feature being enabled by default, promising that the data would be "immediately destroyed."

According to China Newsweek, Taiyuan Chengming Technology pointed out that upload records still existed in the early hours of the day of the apology and questioned whether the data was transferred overseas to a Singapore entity, subsequently sending a letter to pursue accountability.

Nearly 400 developers formed a rights protection group, and the incident escalated from a technical vulnerability into a data compliance and trust crisis.

Facing continued pressure, Z.AI announced on the evening of September 20 that its MaaS platform would launch a "no data content retention" feature and promised to open-source ZCode and invite the China Academy of Information and Communications Technology to conduct a security audit.

On September 21, it was learned that ZCode, the AI programming tool under Chinese large model company Z.AI, was about to be officially open-sourced.

Z.AI stated that this open-sourcing was the first step in responding to external concerns about ZCode's data handling; in the future, it will also push ZCode to move more thoroughly toward community-driven development, welcoming outsiders to download and inspect the code and jointly participate in improvements.

Regarding security, Z.AI revealed that it has invited the China Academy of Information and Communications Technology to conduct a security audit of ZCode, focusing on verifying data retention.

It will also establish a normalized product security vulnerability reporting mechanism in the future, welcoming developer partners to jointly inspect and provide feedback on issues, and the platform will give corresponding rewards based on the severity level of the problems.