PingCAP published a security incident report on 19 October 2026 saying that between 9 September and 7 October 2026 an unauthorised party used stolen service-account credentials to access certain PingCAP internal systems, copy data and obtain source code.

PingCAP says it became aware on 7 October and began disabling the affected credentials that day, deleted further compromised credentials and removed excessive cross-project permissions by 10 October, and that as of 14 October all identified compromised credentials were invalidated and identified unauthorised access paths blocked.

PingCAP says exported records included TiDB Cloud user names and email addresses, organisation and project associations, SQL editor text and errors, and some organisation, project and cluster identifiers, plus business and billing information (customer contacts, contract terms, orders, pricing, payment terms and billing configuration); it says this does not mean every customer had every category exposed and it is still establishing per-customer impact.

It says an exported table contained encrypted TiDB Cloud API-key values and that one investigator-owned value tested by its external incident response firm could not be used directly as an API key, while whether other usable credentials were exposed is still being determined.

PingCAP says the external firm found no evidence the attacker ran commands or kept persistent access in the production systems reviewed, and that with outside counsel and forensic specialists there is no evidence data stored in TiDB was impacted.

It says the firm reported four vulnerabilities on 12 to 13 October, which PingCAP fixed by 14 October, and that it is checking production logs for earlier exploitation.

PingCAP says it will notify affected customers directly.

Source: PingCAP security incident report (19 Oct 2026).

UPDATE (24 Oct 2026): PingCAP co-founder and CEO Max Liu published a note on 23 October apologising to customers, saying 'This happened on my watch' and that he takes responsibility for it, and pointing to a completed independent investigation by NebuSec.

NebuSec's report (findings as of 21 October) says the attacker got in through a compromised GitHub account exposing private repositories, then used keys and tokens harvested from those repositories and Terraform state to reach Google Cloud, Jira and the internal chat system used for TiDB alerts. Through Jira support tickets it harvested credentials customers had shared during support work, including AWS keys that reached several S3 buckets holding customers' TiDB diagnostic data.

NebuSec says eight compromised Google Cloud service accounts were used, 6,458 BigQuery tables were exported on 23 and 24 September and Cloud Storage data was downloaded with rclone; the data taken includes TiDB Cloud control-plane metadata, customer SQL editor history, AI interactions and gateway connection logs, CRM and billing records, support tickets and attachments with embedded credentials, source code, Jira issues and internal chat history. NebuSec found no evidence that data in customer TiDB databases was accessed or of attacker command execution or persistence in production, and says the encrypted TiDB Cloud API-key values could not be decrypted because the encryption key was not exposed, which does not apply to credentials shared in support materials.

It says the four vulnerabilities it found were patched and retested with no evidence of earlier exploitation, and it publishes six attacker IP addresses. (source: PingCAP CEO note; NebuSec)