OpenCode, an open-source AI coding agent, said attackers breached an older system tied to OpenCode Console, affecting 1.4% of Console users and exposing some of the API keys they used for AI services.
Here's what to know
According to TechCrunch, OpenCode emailed affected Console users after attackers accessed the online platform where some people store keys in workspaces.
OpenCode said the incident did not affect its command-line tool or its open-source app.
On X, co-founder Dax Raad wrote, "Note this is the OpenCode Console, not OpenCode itself, those keys never come to us."
Developer posts followed quickly, describing spending alerts and unauthorized activity across OpenAI, Anthropic, and Google AI accounts. Reported losses ranged from a few hundred dollars to more than $1,000.
Although OpenCode called the incident a "large-scale distillation attack," some observers questioned whether the available evidence actually shows the breach was used for model distillation.
TCD Picks » Quince Spotlight
💡These best-sellers from Quince deliver affordable, sustainable luxury for all
The company still hasn't said how attackers accessed the older system, how the keys were stored, or the full extent of the misuse, per TechCrunch.
More background
Raad said attackers had been hitting OpenCode's systems for 48 hours, forcing the team to keep AI defensive agents running continuously.
He also said some models were initially useful in analyzing the intrusion before later refusing to help once the malicious code triggered their safety restrictions.
That issue appeared to ease after OpenCode gained access to OpenAI's Daybreak Blue program, which offers vetted users models less likely to reject defensive cybersecurity work. Raad said that access improved the team's ability to investigate the attack.
TechCrunch reported that, in July, Hugging Face said commercial AI models blocked analysis of logs from an intrusion by autonomous AI agents, writing that guardrails "cannot distinguish an incident responder from an attacker."
The breach showed how closely AI is becoming tied to broader infrastructure demands.
When those tools are misused or fail during an attack, the consequences can extend beyond one company, bringing higher costs, greater security concerns, and additional strain that reaches beyond a single platform.
What's being done?
Anthropic and Google disabled API keys they believed had been exposed. OpenAI said it worked with OpenCode to restore access and reimburse API customers for credits used in the attack.
OpenCode said it is processing refunds for unauthorized charges and advised affected users to revoke those keys and replace them directly with each provider.
So far, the most immediate protective step the company has described for customers who stored keys in Console workspaces is rotating those credentials with each provider.
Access to Daybreak Blue gave the company a model better suited to security response, and vetted pathways for incident responders could become increasingly important as AI-assisted attacks become more common.
Where can I learn more?
The OpenCode breach comes as concerns grow over how quickly AI systems are being built, deployed, and secured. The articles here dig into the security backlash around AI, a leaked Anthropic codebase, and the spending and infrastructure strains tied to the boom.
• Public anger over AI's growing security risks intensified as power and water demands kept climbing.
• Anthropic moved to contain leaked code used to direct one of its AI agents.
• Analysts said AI data center growth is driving heavier electricity and water use.
• Big Tech developers used lucrative land deals to secure space for new data centers.
Get TCD's free newsletters for easy tips, smart advice, and a chance to earn $5,000 toward home upgrades. To see more stories like this one, change your Google preferences here.
