Command Code revokes all API keys in response to a suspected AI distillation attack targeting model keys

Article

After OpenCode and Railway, Command Code has also come under attack: hackers are targeting AI model keys

Dongcha Beating AI breaking news,October 29，AI coding tool Command Code announced that it had blocked a security incident related to a large-scale distillation attack targeting Anthropic and OpenAI models. The company proactively revoked all of its own API keys, and users need to log in again to generate new keys.

Over the past 2 week, OpenCode and Railway successively suffered similar attacks.October 17, OpenCode's cloud Console was compromised, affecting approximately 1.4%of Console users. Some third-party model keys may have been exposed, and users reported hundreds or even thousands of USD。

October 21, Railway was also breached, with up to approximately 100000 workspaces at risk of environment-variable exposure. Investigators found in the attacker's code logs that the hackers were specifically searching for AI service's API keys. Railway determined that the attackers were likely attempting to use these keys to make large numbers of model calls and collect the outputs for distillation.

Earlier,October 14 Zeabur was also breached. The attackers focused on searching for AI API keys and even attempted to use the stolen AWS credentials to call Amazon Bedrock, but were unsuccessful.

Original link https://m.theblockbeats.info/flash/374338

Key points

Command Code announced that it had blocked a security incident related to a large-scale distillation attack targeting Anthropic and OpenAI models.
Command Code preemptively revoked all of its own API keys, and users need to log in again to generate new keys.
Over the past 2 week, OpenCode and Railway experienced similar attacks in succession.
The OpenCode Cloud Console was compromised, affecting approximately 1.4% of Console users; some third-party model keys may have been exposed.
Railway had approximately 100000 workspaces at risk of environment-variable exposure, and the attacker was found to be specifically searching for AI service API keys.
Zeabur was previously compromised, and the attacker attempted to use stolen AWS credentials to call Amazon Bedrock, but was unsuccessful.

AI-assisted interpretation

The following is analysis, separate from reported facts. Verify important claims independently.

This is a report about the security of AI tool keys. Attackers may search for API keys, then use them to call models and collect outputs; Command Code said it revoked its own keys, and users must log in again to obtain new keys.

Why it matters to readers

API Once exposed, keys may be used by others to call models and incur charges, or may be used to collect model outputs.

If you use AI programming tools or third-party model services, you should pay attention to key rotation, re-login, and subsequent official guidance.

Risks and unknowns

The provided materials do not include a Command Code announcement or details of an independent technical investigation.
The materials do not state the scope affected by the Command Code incident, the attack vector, or whether there was any misuse of user keys.
The report does not state the final investigation findings regarding the OpenCode, Railway, and Zeabur incidents.

Command Code revokes all API keys in response to a suspected AI distillation attack targeting model keys

Article

AI-assisted interpretation

Why it matters to readers

Risks and unknowns

Related Developments

Related concepts

Keep following

Security Center