There are reports of a potential API key leak involving the AI programming tool 'OpenCode,' and that the account balances for OpenAI and Claude have been wiped out due to fraudulent use.

The operators of the open-source AI programming tool 'OpenCode' announced on October 17, 2026, that their system had been compromised. API keys for using AI services such as OpenAI and Anthropic may have been leaked, and some users have reported being charged for AI services they did not use.
We had a security breach last night related to a large scale distillation attack that impacted 1.4% of OpenCode Console users
— OpenCode (@opencode) October 17, 2026
We have notified them all this morning via email. If you did not get an email you were not impacted, it was a subset of users who had external providers…
OpenCode is a tool that automates program creation and modification by giving instructions to AI. It supports multiple AI models, including OpenAI's GPT, Anthropic's Claude, and Google's Gemini, allowing users to choose and use their preferred AI model.
OpenCode, a free and open-source AI coding agent, is available for Windows, Linux, and macOS, and supports Claude, GPT, Gemini, and more - GIGAZINE

One way to use external AI models is by registering an 'API key.' An API key is authentication information used to identify service users, and if it is leaked to a third party, there is a risk that the AI model will be misused using the owner's account, which could lead to high charges depending on usage.
Anomaly Innovations, the company that operates OpenCode, has notified users that it has detected unauthorized access to its old system, which was being decommissioned. Users of the 'OpenCode Console,' an online service used for connecting with external AI services, are affected. According to the notification, API keys for OpenAI, Anthropic, and other services stored in users' workspaces may have been leaked.
OpenCode explained on its official X account that 'a security breach related to a large-scale distillation attack occurred, affecting 1.4% of OpenCode Console users.' They stated that all affected users have been notified via email.
It should be noted that OpenCode Console is a separate service from the regular OpenCode app. This data breach affected some users who had linked external AI services to OpenCode Console, but users who only use the regular OpenCode app are not affected.
On X, some users have already come forward to report being victims. Max Weinbach, who publishes IT-related information, reported that 'my balances for OpenAI and Anthropic were depleted, and my Google account was restricted and some of the balance was also consumed.' He also received an email from OpenCode informing him of the data breach.
Can confirm
— Max Weinbach (@mweinbach) October 17, 2026
My OpenAI and Anthropic balances got drained, Google account got limited and drained a bit and OpenCode did email me https://t.co/HxPrmXJe2U pic.twitter.com/sdR8D1OAq2
According to the news site Notebookcheck , there have been reports of users incurring unexpected charges of $400 (approximately 60,000 yen) on their Anthropic accounts, and of users incurring charges of approximately 200 euros (approximately 35,000 yen) on Google services, resulting in their projects being suspended. However, a causal relationship between these individual incidents and the recent unauthorized access has not been confirmed.
To prevent further damage, Anthropic has also stepped in to address the issue. On October 18, 2026, Lydia Harry of Anthropic reported that they were working with OpenCode to address the problem and had already invalidated the affected Claude API keys. Users who had stored their API keys with OpenCode are being urged to create new API keys as a precaution.
We've been working with OpenCode on yesterday's breach and have revoked Claude API keys we believe were affected.
— Lydia Hallie ✨ (@lydiahallie) October 17, 2026
If you saved a key in OpenCode and want to be extra careful, please create a new one! https://t.co/25s4VX1rfw
Furthermore, OpenCode developer Dax posted on October 18th that 'attackers have been continuously attempting to attack the system for the past 48 hours.' While the system's security has been strengthened in the process of dealing with the attacks, he stated that the attackers are running AI agents 24/7, and that the defense needs to have a similar system in place.
they've been non stop trying to attack our systems for the past 48 hours
— dax (@thdxr) October 18, 2026
a side effect is everything is getting hardened but damn we basically need agents going 24/7
Because that's what they have
OpenCode is asking affected users to invalidate and reissue their API keys and to review their usage history and billing details. OpenCode is also working with Anthropic, OpenAI, and Google to resolve the issue, but has stated that 'we cannot share details because the problem is ongoing,' and the full extent of the intrusion and damage will likely be revealed in future investigations.
Related Posts:
in AI, Web Service, Security, Posted by log1d_ts







