Article

Breaches Now Start With the Credentials You Forgot

Three recent incidents point to the same weakness: stolen or exposed credentials and keys, sitting in plain sight, doing more damage than any exploit.

HemeswariOctober 25, 20264 min read

The common thread in the recent incidents logged on this beat is not a novel exploit. It is credentials and keys that were already sitting inside systems, exposed or stored badly, and then used against the organizations that held them. Zeabur, CISA's accounting of SafePay's victims, and the OpenCode exposure all describe breaches that turn on access material rather than on a clever technical break-in.

Plaintext Is the Whole Story at Zeabur

The clearest case is Zeabur, the development platform that is shutting down after its second security breach in two months, as BleepingComputer reported. The disclosed incident involved an attacker using leaked internal credentials that had been stored in plaintext to compromise infrastructure. The haul was large and unusually specific: more than 3.2 million environment variable records and information for 321,810 accounts.

That detail matters because environment variables are where modern software keeps its secrets. A development platform holds exactly the material its customers use to reach databases, cloud services and third-party APIs. Storing those credentials in plaintext turns an internal compromise into a supply-chain event, and the account count shows how quickly a single set of internal keys becomes a customer-facing breach. The second breach in two months also suggests remediation did not address how the credentials were held in the first place, or that the same exposure survived the first cleanup.

SafePay Shows the Scale of Credential-Driven Intrusions

The CISA advisory on SafePay, also reported by BleepingComputer, puts the pattern at infrastructure scale. The agency says the SafePay ransomware gang has breached over 600 critical infrastructure organizations worldwide since 2024. A ransomware operation of that size depends on repeatable access, and the recurring path into organizations is stolen or weak credentials rather than bespoke tooling. The 600-plus figure is a measure of how many organizations had access material that could be bought, guessed or reused.

For US technology companies, the relevant point is that critical infrastructure is now heavily software-defined. The operators on CISA's list depend on cloud platforms, managed services and API-driven vendors, many of them American. A gang that gets in through credentials at an operator can reach the vendors connected to it, and vice versa. The SafePay numbers describe an access problem that crosses the boundary between a customer and its suppliers.

OpenCode Exposes Keys, Not Just Accounts

The third incident involves the open-source AI coding project OpenCode, which notified a small number of users that API keys they used with external providers were exposed in a security breach, per BleepingComputer. The affected population is small, and the notification language suggests the project moved to contain it. But the asset that leaked was not a password to a discussion forum. It was a working key to an external provider, which means the exposure lands on third parties who never had a relationship with OpenCode's users.

That is the same structural issue as Zeabur, at a different scale. AI coding tools chain together providers, and the keys that connect them are the weak link. When a project stores or transmits those keys in a way that fails, the damage surfaces in billing, in provider-side abuse and in the uncertainty of not knowing which key was used where. The small user count understates the reach of the disclosure.

What This Means for US Companies and Consumers

Taken together, the three stories describe a shift in where breach risk sits. Organizations have spent years hardening the perimeter and patching exploitable software. The incidents above did not require either. They required only that access material be present, stored in a form an attacker could read, and reused across systems.

For US technology companies, the practical consequence is that secret management is now a breach-control function, not a hygiene checkbox. Zeabur shows what happens when internal credentials sit in plaintext at a platform that holds customer secrets. OpenCode shows what happens when API keys escape their intended boundary. Neither event required an advanced adversary.

For US consumers, the effect is indirect but real. When a development platform with 321,810 accounts is compromised, the downstream services those accounts touch inherit the risk. When API keys are exposed, the provider's abuse controls and billing systems absorb the cost, and those costs eventually reach customers. The SafePay figures show that this is not confined to startups: critical infrastructure operators are in the same pool.

The Uncomfortable Part Is the Remediation

A credential-driven breach is difficult to close because the fix is operational. It requires inventorying where secrets live, rotating them, and confirming that the old material is dead. The Zeabur sequence, two breaches in two months, suggests how hard that is in practice. Rotating one set of keys is straightforward; proving that no plaintext copy remains in logs, configuration or backups is not.

The same difficulty applies to API keys handed to external providers. Revocation is only as good as the provider's process for honoring it. OpenCode's notification to a small number of users is the right first step, but the useful measure is whether the exposed keys were invalidated everywhere they could be used.

What to Watch

Three signals are worth tracking against these stories. First, whether Zeabur's shutdown includes a clear account of how the plaintext credentials were stored and for how long, which would set a benchmark for other development platforms. Second, whether CISA's SafePay advisory is followed by sector-specific guidance on credential controls at critical infrastructure operators, since the 600-plus figure implies an access problem that general ransomware advice has not addressed. Third, whether OpenCode's disclosure is followed by evidence that the exposed keys were revoked with the external providers, which would show whether the small-notification model works for API key incidents.

The pattern itself is unlikely to change. Credentials and keys will keep sitting inside systems, and attackers will keep using them. What can change is how quickly organizations know where that material lives and whether it still works.

More on this beat: Cybersecurity on TechManNews.

#data breaches#credentials#API keys#ransomware#critical infrastructure#secret management

Newsletter

Get Tech News in Your Inbox

The latest AI, gadgets, software and startup stories from TechManNews, delivered every morning - free.