"Action Required: Potential provider key breach." That was the subject line OpenCode sent on the morning of October 17 to customers it says may have had third‑party AI provider keys exposed after an intruder reached a legacy system on the night of October 16.
OpenCode's timeline and notice to users
OpenCode says an intruder reached a legacy system at OpenCode on the night of October 16 and that the incident involved OpenCode Console, the company's optional hosted workspace service. The company emailed affected customers on the morning of October 17; anyone who did not receive that message, OpenCode said in a post on X the same day, was not affected. OpenCode characterized the incident as ongoing on October 17 and said it was working with Anthropic, OpenAI and Google to fix it.
Console‑stored provider keys and what was exposed
Console saves a workspace's OpenAI, Anthropic, or Google API key and uses that key to call the provider so team members can use models without seeing the key. OpenCode said the affected group is 1.4% of Console users, all of whom had an outside provider linked. OpenCode co‑founder Dax Raad wrote on X that the breach involves "the OpenCode Console, not OpenCode itself," and added that those keys "never come to us." The company has described the breach as "related to a large-scale distillation attack," but has not named who was responsible or provided technical details on the legacy system, how it was reached, how many keys it held, or how many individual users the 1.4% represents.
This site is the portfolio.
OSINTSights runs on Cloudflare Workers, D1, R2, and Vectorize, with an AI pipeline on Hetzner ARM. Nubivance designed, built, and operates it. We do the same for clients.
See what we buildProvider‑recommended remediation steps for affected accounts
- Revoke the key in the provider's dashboard and create a new one, following OpenAI, Anthropic, or Google's instructions.
- Update every application and script that used the old key.
- Review usage and billing at the provider for requests you did not make, going back at least to October 16.
- Set a spend limit or alert at the provider and give each tool its own key so a copied key does less damage and its use is easier to trace.
OpenCode also warned that deleting a key from Console does not revoke it, and that a Console budget cannot cap misuse because Console only sees requests that pass through its own gateway. Console keys pasted into the hosted workspace are the ones OpenCode identified as at risk; by contrast, the OpenCode command‑line and desktop apps keep provider keys locally in a SQLite database on the user's machine.
Reported unauthorized charges tied to exposed keys
OpenCode framed the keys as ones that "may have been exposed" and has not said that any key was definitively copied or used. Still, several users reported billing they attribute to the incident. Dealroom reported that Colin Parsons wrote on X that his OpenAI key had been used for $281 and his Anthropic key for $168, and that Preston Zeller put his losses at $1,700. Notebookcheck published Reddit posts claiming about $400 of unauthorized Anthropic usage and around €200 of Google usage, the latter followed by account suspension. These are the users' own accounts and assertions; OpenCode has not confirmed copied keys or unauthorized provider charges companywide.
Distillation attacks, Anthropic's September advisory, and continuing intrusion attempts
OpenCode described the incident as related to a "large-scale distillation attack." Distillation is the industry term for campaigns that pull a model's answers in huge numbers to train rival models. Anthropic had warned in September that the labs behind such campaigns reach its models via networks of fake accounts created with stolen credit cards, login credentials and harvested API keys. OpenCode said attackers were still trying to access its systems on October 18; Raad wrote on X that attackers had been "non stop trying to attack our systems for the past 48 hours," and that "a side effect is everything is getting hardened."
What this means for technologists, procurement leaders, and end users
- Technologists and security teams: expect to revoke and rotate any keys stored in Console, review provider billing for anomalous usage back to October 16, and prefer per‑tool keys with provider spend limits so a single exposed key is easier to trace and contain.
- Procurement and enterprise IT leaders: note that OpenCode said the affected set was 1.4% of Console users and that Console budgets do not limit provider billing; teams will want inventory of which applications share provider keys and require separate, auditable tokens for distinct tools.
- End users and individual developers: follow provider guidance to revoke and replace exposed keys, check provider dashboards for unauthorized charges, and be aware that deleting a key from Console does not itself invalidate the credential.
OpenCode is working with the named providers and has signaled continued intrusion attempts and hardening activity. The company has not published technical forensic details — including when unauthorized access may have begun, how the legacy system was reached, or the total number of keys held — leaving those concrete follow‑ups to come from OpenCode and the providers as they coordinate remediation and investigation.
Source: The Hacker News — OpenCode Console Breach May Have Exposed OpenAI, Anthropic, and Google API Keys
