On October 17, 2026, it was revealed that the popular open-source AI programming tool OpenCode had been compromised. The information was published by Livedoor.com, which reported that API keys needed to access OpenAI, Anthropic (Claude), and other AI services may have been stolen during the incident. While no official statement has been made confirming that the keys were leaked, some users have reported unexplained AI usage charges that have reduced their account balances to zero.
OpenCode is an open source project that allows developers to integrate various generative artificial intelligence (AI) services directly into their workspaces. The tool uses API keys to access both the OpenAI and Anthropic platforms. When these keys fall into the wrong hands, it is possible to potentially use AI models without the user’s consent, creating an unwanted increase in resource consumption.
Description of the incident and confirmed facts
A report from Livedoor.com states that the OpenCode administration suffered a „large-scale distillation attack“ affecting around 1.4 million OpenCode Console users. The attack allegedly allowed hackers to access internal data, including API keys used for OpenAI and Anthropic services. While the source does not provide an exact number of stolen keys, it does emphasize that this is a „potential“ leak.
User feedback collected by Livedoor.com indicates that some users have noticed their account balances for AI services suddenly dropping to zero. These reports indicate that the unfamiliar AI request charges may have been related to leaked credentials. While the source does not confirm a direct connection between the hack and the specific charges, it does provide context for the potential unauthorized use of AI.
Related tools and additional information
Another source, pypi.org, lists the package „mcp-openai-pro 2026.10.19.5,“ which is „MCP Server for OpenAI API via AceDataCloud.“ While this post doesn’t provide direct evidence of a breach, it does show that OpenAI API integrations are widely used across platforms, including third-party servers. This highlights that protecting API keys is critical not only for core tools but also for components of their ecosystem.
Security challenges and possible solutions
In the case of OpenCode, it is important to note how distillation attacks can affect even well-audited systems. Distillation is a method where an attacker attempts to extract internal data from model responses, and a large-scale attack can escalate this process to a credential leak. This threat is especially relevant when API keys provide direct access to paid AI models.
To mitigate risk, tool developers should implement multi-factor authentication, rotate API keys regularly, and use rate limiting and a monitoring system that can quickly detect unusual usage patterns. Users are also advised to periodically review their API key usage statistics and, if possible, use separate keys for different parts of the project.
What we know so far and what we still need to watch
Until this announcement, OpenCode was considered a reliable tool for developers to quickly experiment with AI models. This disclosure of the security breach shows that even well-known projects can become targets. While there is currently no confirmed data on widespread losses or significant financial impact, users should remain vigilant and monitor their account balances.
Conclusions
The OpenCode security breach is an important reminder of the importance of API keys and the potential consequences when they fall into the wrong hands. While sources have not confirmed specific losses, user reports of unexpected charges for AI services indicate that the situation requires attention. Security practices such as key rotation, multi-factor authentication, and constant usage monitoring can help reduce the risk of similar incidents in the future.
Sources
- Livedoor.com - AI programming tools「OpenCode」でAPIキー流出のとなたとたととさい
- Pypi.org - mcp-openai-pro 2026.10.19.5
