Google is amplifying its efforts in the realm of cybersecurity. During its I/O event, the tech giant revealed it is inviting select specialists to trial the API for CodeMender, an “AI agent for code security” that was initially introduced last October. The significant shift is that Google is now expanding the tool’s accessibility to a broader audience. The company is promoting it, as Google DeepMind CTO Koray Kavukcuoglu noted, as a solution to “help secure the world’s code bases” by identifying and remedying vulnerabilities.
Anthropic’s unexpected announcement regarding the Claude Mythos Preview caught the AI industry off guard, affecting key sectors such as major banks and even the Federal Reserve. In response to Anthropic’s move, numerous companies have been inspired to develop their own sophisticated AI models aimed at uncovering hidden security flaws in critical systems. The Mythos Preview gained attention for being perceived as too potent for a public release, aiding Anthropic in mending its relationship with the US government following its supply chain risk designation and subsequent legal challenges. If successful with its initial enterprise and government users, the initiative could be financially rewarding for Anthropic. With mounting pressure for companies to show profitability before potential IPOs, like OpenAI, or maintain a competitive edge in the accelerating AI race, as Google aims to, cybersecurity is emerging as a promising revenue stream for AI labs.
OpenAI was quick to introduce its own version, and now Google has joined the fray. Kavukcuoglu confirmed in an interview with The Verge that Google has been engaging in conversations with both governmental bodies and enterprises about leveraging CodeMender for auditing their systems.
During a press briefing on Monday, Google CEO Sundar Pichai acknowledged Anthropic’s achievement, saying, “What Mythos has done, and credit to them, is to demonstrate the value of large-scale models in security applications. However, I believe this is an area where we also have strong capabilities.”
