The OpenCode API breach has exposed stored credentials belonging to some users who connected external AI providers to its Console. Affected developers reported unauthorized usage charges, while others said Google suspended their cloud projects after detecting suspicious activity.
According to OpenCode, the incident affected 1.4% of Console users. The project linked the breach to a large-scale AI distillation attack and said it had emailed everyone it identified as affected.
Some users reported unexpected bills of around $1,700 or more overnight. However, the total financial impact remains unknown.
Breach Affected Users With External Providers Connected
OpenCode is an open-source AI coding agent that helps developers understand software, build features and fix bugs through natural-language instructions. Users can access it through a terminal, desktop application or code editor.
The service also supports models from multiple providers. Developers can connect external accounts to the OpenCode Console, including services from Anthropic, OpenAI and Google.
According to the project’s announcement, the breach affected only a subset of users with those external connections. OpenCode said users who did not receive a notification email were outside the affected group.
Its notice identified October 17, 2026, at 00:30 UTC as the earliest possible exposure time for the keys.
OpenCode Links Incident to AI Distillation Activity
OpenCode described the incident as a large-scale distillation attack. Distillation normally involves training a smaller AI model using outputs from a more capable model.
The technique can help developers create faster and cheaper systems. However, unauthorized access to AI services can turn that process into abuse, particularly when attackers generate large volumes of responses using stolen credentials.
In this case, OpenCode has not disclosed the full technical sequence behind the breach. Consequently, questions remain about how attackers obtained the keys and how much unauthorized activity they carried out.
Developers Report Unexpected Bills and Project Suspensions
Following the OpenCode API breach, users reported charges ranging from several hundred dollars to roughly $1,700 or more. These accounts describe unauthorized API usage after the credential exposure.
Meanwhile, some developers said Google suspended their projects after detecting activity consistent with hijacked resources.
Such suspensions can interrupt legitimate work alongside the unauthorized usage. However, the available reports do not establish how many projects Google suspended.
OpenCode’s notification mentioned possible reimbursement for affected users. It did not provide a confirmed payment schedule or detailed eligibility terms in the information available.
Co-Founder Describes Sustained Attacks Using AI Agents
OpenCode co-founder Dax Raad said attackers had targeted the company’s systems for at least 48 hours using AI agents.
He also described difficulties during the defensive investigation. According to Raad, the AI models his team used began refusing requests because of their built-in safeguards.
Raad said the situation stabilized after OpenCode gained access to OpenAI’s Daybreak cybersecurity model for defensive work on Saturday.
These statements describe the team’s response, but they do not provide a complete account of the intrusion or its underlying cause.
Investigation Continues With AI Providers
OpenCode said it was working with Google, OpenAI and Anthropic to address the incident. However, it had not released further technical details at the time of the report.
Although the company confirmed API key exposure, its notification said no data appeared to have been exposed. The available information does not clarify the scope of that statement or establish exposure of users’ code, prompts or other content.
For now, the confirmed disclosure concerns stored provider credentials belonging to a limited share of Console users. User reports also indicate financial and operational consequences, while the investigation continues.
0 responses to “OpenCode API Breach Exposes Keys as Users Report Unauthorized Charges”