Google Aims to Rival Anthropic’s Mythos

Google is ramping up its efforts in the cybersecurity arena. At its recent I/O event, the company revealed that it is inviting select experts to test the API for CodeMender, an “AI agent for code security” introduced last October. This time, Google is expanding access to the tool, promoting it as a means to “help secure the world’s code bases,” as stated by Google DeepMind CTO Koray Kavukcuoglu, by identifying and addressing vulnerabilities.

The AI community was taken by surprise with Anthropic’s announcement of the Claude Mythos Preview, which caught the attention of major players, including top banks and the Federal Reserve chair. Following Anthropic’s lead, numerous companies are now launching their own advanced AI models aimed at uncovering hidden security flaws in critical systems. Mythos Preview gained notoriety for being deemed too powerful for public release, which has helped Anthropic regain favor with the US government after facing supply chain risk issues and a subsequent lawsuit. If successful with its early-access enterprise clients and government partnerships, the company stands to generate significant revenue. As the pressure mounts for firms to turn profits ahead of potential IPOs, like OpenAI, or to maintain their edge in the competitive AI landscape, such as Google, AI labs are increasingly viewing cybersecurity as a vital revenue stream.

OpenAI quickly responded with its own solution, and now Google has followed suit. Kavukcuoglu confirmed in an interview that Google is already in talks with governments and enterprises about utilizing CodeMender for system audits.

During a press briefing, Google CEO Sundar Pichai acknowledged the impact of Mythos, 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.”