Critical x402 SDK Flaw Could Allow Malicious APIs to Drain AI Agents' Token Balances

On October 22, 2026, the V12 security firm reported a critical flaw in x402 SDKs that could allow malicious APIs to drain AI agents' full token balances. The vulnerability arises from the SDKs' default client, which originally set no spending limit for API payment requests. According to the report, releases issued on August 18, 2026, added spending limit functionality to mitigate the issue. However, the risk remains significant for AI agents that automatically sign payments and whose applications have not been configured with explicit spending limits. The flaw affects x402 SDK integrations across AI agent workflows, where a malicious API endpoint could trick the SDK into signing a payment for the agent's entire supported token balance. V12 noted that the exploitation is limited to agents with automatic payment signing and insufficient spending limit configurations. The firm recommended that developers update to the latest SDK versions and configure explicit spending limits to mitigate the risk. As of the disclosure, no CVE identifier had been assigned to the vulnerability.