Google’s Surge in Chrome Vulnerability Discoveries Likely Driven by AI

The number of Chrome vulnerabilities discovered by Google has surged over the past month, likely driven by the company’s use of AI.

Chrome security advisories published by Google in late March and early April mentioned a handful of vulnerabilities “reported by Google”, but the number increased to 16 for the Chrome update released on April 15 and 21 for the update issued on April 28.

The number of vulnerabilities found by Google surged to 100 in the advisory published on May 5. More than 70 vulnerabilities patched in the two most recent Chrome releases were found internally by the tech giant.

While Google has not specifically said the vulnerabilities were discovered using AI, the timing of the surge and other aspects suggest that artificial intelligence is responsible.

When it recently announced lowering Chrome bug bounties, Google noted that AI and automation have been helping its teams move “at an unprecedented rate – remediating risks more effectively than ever before.”

“The latest advancements in AI from Google and the broader industry have made it significantly easier to take a test case and explain the root cause, propose a suitable fix, and to find variants of known problems,” Google said.

Other major organizations have also reported surges in vulnerability discovery thanks to the use of third-party or internal AI tools.

Mozilla, for instance, found over 270 Firefox vulnerabilities with the aid of Anthropic’s new Claude Mythos model. Microsoft and Palo Alto Networks have also found many vulnerabilities in their products using advanced AI vulnerability discovery tools.

It’s unclear which AI model may have been used by Google to discover the Chrome vulnerabilities. Google is one of roughly 50 organizations granted access to Claude Mythos.

Google has also been working on its own AI-powered vulnerability discovery tools, such as Big Sleep and CodeMender.

“CodeMender is an AI code security agent, originally developed by Google DeepMind. Leveraging Agent Platform capabilities and advanced Gemini models, CodeMender autonomously identifies vulnerabilities within your code,” Google said in a blog post this week.

“It then recommends precise fixes, securely tests them, and can apply patches and necessary changes across dependent systems, with your approval. This entire process automates secure deployment while ensuring your developers retain control,” it added.

The tech giant may have leveraged AI tools that are used only internally.

Google has not responded to SecurityWeek’s questions regarding the exact number of vulnerabilities discovered by AI in Chrome and which model or tool has been used to find the flaws.

This article was published by Security Week. Please check their website for the original content.