Braintrust, the AI‑evaluation startup that positions its platform as an operating system for engineers building artificial‑intelligence software, confirmed an unauthorized access event in one of its Amazon Web Services (AWS) accounts. The breach, revealed in an email to customers on Monday and posted on the company’s website Tuesday, involved API keys that clients store with Braintrust to connect to cloud‑based AI models.

According to the notice, Braintrust identified the intrusion, locked down the compromised account, and conducted a full audit of related systems. The company also rotated its own internal secrets and urged every customer to "rotate any of the API keys that they store with Braintrust." While the firm has spoken with one impacted client, it says there is no evidence that the exposure extends beyond that case.

Martin Bergman, a spokesperson for Braintrust, told TechCrunch the email was sent "out of an abundance of caution." He emphasized that the company has confirmed a security incident but has not found proof of data being stolen or used maliciously. The root cause of the breach remains under investigation.

Company response and impact

Braintrust’s platform enables enterprises to monitor AI models and products, a service increasingly vital as more firms embed machine‑learning components in their operations. Founder and CEO Ankur Goyal has described the tool as akin to an "operating system for engineers building AI software." The startup recently secured $80 million in a Series B round that valued it at $800 million, underscoring its rapid growth and the trust placed in its technology.

Jaime Blasco, co‑founder of cybersecurity firm Nudge Security, received the breach alert and warned that the incident could have downstream implications for customers that rely on Braintrust’s services. If hackers obtain API keys, they can impersonate legitimate users and potentially access a client’s broader cloud environment without breaching the client’s own defenses.

Broader cloud‑security context

The Braintrust episode joins a series of high‑profile cloud‑account compromises. In 2023, CircleCI disclosed a similar breach in which attackers harvested stored secrets, prompting the company to advise users to rotate all credentials. More recently, an EU cybersecurity agency reported that hackers stole 92 gigabytes of data from a compromised AWS account used by the European Commission, affecting dozens of internal clients.

Experts note that cloud platforms are attractive targets because a single stolen API key can grant extensive access across multiple services. Organizations are therefore urged to adopt strict key‑management practices, including regular rotation, least‑privilege policies, and continuous monitoring for anomalous activity.

Braintrust says it has taken immediate steps to mitigate risk and is working with security researchers to understand how the breach occurred. Customers are encouraged to follow the rotation instructions promptly and to review any unusual activity linked to their API keys.

Used: News Factory APP - news discovery and automation - ChatGPT for Business