Google Takes On Anthropic’s Mythos

Google is significantly intensifying its focus on cybersecurity, unveiling broader access to its CodeMender tool. This "AI agent for code security," first introduced last October, is now being made more widely available externally, with select groups of experts invited to test its API. Google is positioning CodeMender as a critical solution to "help secure the world’s code bases," capable of both identifying and remediating vulnerabilities, as articulated by Google DeepMind CTO Koray Kavukcuoglu.

This strategic move by Google comes amidst a wave of activity in the AI security sector, largely spurred by Anthropic’s surprising Claude Mythos Preview announcement. This powerful AI model, designed to uncover hidden security gaps in critical systems, reportedly sent shockwaves through the AI community, including top financial institutions and the Federal Reserve. Anthropic garnered significant attention for Mythos Preview's alleged power, deemed too potent for public release, a development that also helped the company mend relations with the U.S. government following a supply chain risk designation and subsequent lawsuit. The model holds substantial revenue potential from its early-access enterprise and government clients. Consequently, as companies like OpenAI face pressure to demonstrate profitability ahead of potential IPOs, and Google strives to maintain its lead in the escalating AI race, cybersecurity is emerging as a pivotal revenue driver for leading AI laboratories.

Following OpenAI's swift introduction of its own security offering, Google has now entered the fray with CodeMender. Kavukcuoglu confirmed in an interview that Google is already engaged in discussions with governments and enterprises regarding the implementation of CodeMender for system audits. Google CEO Sundar Pichai addressed reporters during a Monday press briefing, stating, “What Mythos has done, and credit to them, is to show that there is a value for the largest-sized model in these kinds of security use cases. But I think it’s something we are capable of doing as well.”