Cascadity Audio
← All streams

#llmjacking

Everything tagged llmjacking, across every stream.

0

"Black October": Wiz counts at least nine AI developer platforms breached in a month, with stolen AI keys resold

Wiz connects the dots across a month of separate breach notices and shows how a stolen AI key gets tested, resold and burned through.

Cloud security firm Wiz says at least nine platforms that developers use to build and host AI apps disclosed security incidents in October, among them Fireworks AI, Lightning AI, OpenCode, Railway, Warp and, on October 30, Northflank. It later added Brev (NVIDIA), Trigger.dev and Chainstack. Wiz says no single vulnerability links them, but every public report describes attackers hunting stored environment variables, especially keys that unlock AI services. Zeabur says it will shut down after its incident.

Wiz traced two stolen enterprise keys: each was first tested with a single call, then flooded with Amazon Bedrock requests, about 2.5 million calls in all and likely over $100,000 in usage at list price, with traffic resembling resold coding-agent use. It links the keys to an LLM-resale operation.

Northflank told customers to assume every environment variable was decrypted and to rotate all secrets; users reported their AI keys being abused. It says the attack vector was contained within an hour, that it patched the Linux flaw CVE-2026-98432 across its fleet, and that a root-cause report will follow.

Cascadic Analysis 2
UndertowPrivilege escalation + credential propagation

What hidden risk could pull against this, even if the news is good?

Wiz says no single bug links the nine incidents, but every public report describes attackers going after stored environment variables, and Railway's own account starts with privilege escalation in its build infrastructure. A hosting platform holds its customers' AI keys so their apps and agents can run, which makes one breach a key ring for many companies at once. Northflank's advice to treat every stored variable as decrypted and rotate all of them shows how far a single foothold can reach.

0
UndertowAutonomous cyber capability scales attackers enormously

What hidden risk could pull against this, even if the news is good?

Within 30 minutes of a key being stolen, one test call came from a residential address; then one key saw about 800,000 Bedrock calls in 30 hours and another about 1.7 million in 13. Fireworks, OpenCode and Warp reported signs of an "agent swarm" attack, and Wiz links the keys to resale operations whose traffic looked like coding-agent use. Theft, testing and resale of AI access now run fast enough that a stolen key can rack up more than $100,000 before anyone notices the bill.

0
Rabbit Holes 1