Model Context Protocol (MCP) is facing renewed scrutiny after researchers discovered that its initial release lacked mandatory authentication, creating significant security vulnerabilities. Pynt's research, initially reported by VentureBeat last October, indicated a 92% probability of exploitation when deploying just 10 MCP plug-ins, with even a single plug-in posing a meaningful risk.

The core issue, according to experts, is that MCP was shipped without built-in authentication, a flaw that authorization frameworks introduced six months after its widespread deployment have failed to fully address. Merritt Baer, chief security officer at Enkrypt AI, had previously warned about the dangers of insecure defaults, stating, "MCP is shipping with the same mistake we've seen in every major protocol rollout: insecure defaults. If we don't build authentication and least privilege in from day one, we'll be cleaning up breaches for the next decade."

The emergence of Clawdbot, a viral personal AI assistant that operates entirely on MCP, has amplified the threat. Clawdbot's ability to manage inboxes and write code has made it a popular tool, but its reliance on MCP means that developers who launched it on virtual private servers (VPS) without proper security configurations have inadvertently exposed their companies to the protocol's vulnerabilities.

Itamar Golan, who anticipated these problems, sold his stake in MCP-related ventures before the full extent of the security flaws became widely known. The situation highlights a recurring problem in the tech industry: the rush to market often leads to security being treated as an afterthought. As Baer pointed out, this approach inevitably results in costly and time-consuming cleanups. The current cleanup is already underway, and the challenges are proving to be more significant than initially anticipated. The industry is now grappling with the consequences of prioritizing speed over security, a lesson that many hope will inform future protocol rollouts.