Google Challenges Anthropic with CodeMender for Cybersecurity

Google has recently intensified its efforts in the field of cybersecurity by unveiling the CodeMender API at the I/O event. This tool, described as an "AI agent for code security," was initially introduced last October. Now, Google is expanding its access and commercializing it to help secure global codebases by identifying and fixing vulnerabilities, as explained by Koray Kavukcuoglu, CTO of Google DeepMind, in an interview with The Verge. Google has invited selected expert groups to test this API, underscoring its commitment to enhancing code security.

This initiative from Google comes amid intense competition, particularly following the surprise announcement from Anthropic regarding Claude Mythos Preview. This AI model has made waves in the industry, especially among major banks and the Chair of the Federal Reserve, due to its purported power, deemed too high for public release. This announcement has allowed Anthropic to strengthen its ties with the U.S. government after a risk designation for the supply chain and a lawsuit.

Google's response to this announcement highlights the growing importance of cybersecurity as a potential revenue driver for AI labs. OpenAI also reacted swiftly with its own offering, emphasizing the pressure on companies to generate profits ahead of potential IPOs or to maintain their lead in the AI race.

Sundar Pichai, CEO of Google, acknowledged at a press conference that Mythos had demonstrated the value of large AI models in security use cases. However, he asserted that Google also possesses the necessary capabilities to excel in this area. Furthermore, Kavukcuoglu confirmed that Google has already engaged in discussions with governments and companies regarding the use of CodeMender in auditing their systems.