LLM Proxy Abuse

LLM proxy abuse occurs when someone gains unauthorized access to a large language model service through legitimate user accounts or API credentials, then uses that access for purposes the organization did not authorize or intend. This might involve reselling API access, running computationally expensive tasks at the organization's cost, or using the model for harmful applications like generating fraud or misinformation. For compliance and governance, this matters because it creates financial exposure, potential liability if the model output causes harm, and evidence of inadequate access controls that regulators and auditors will flag as a control failure.

1 item