How Google AI Innovations are Revolutionising Cyber Defence

Cyber threats are growing in sophistication, meaning defenders are on the look out for new tools to outpace these evolving risks. 

AI has both positive and negative impacts in this space: while it greatly improves defences with automation, advanced threat detection and predictive insights, it also brings fresh risks — such as the possibility of malicious use and the emergence of new attack methods.

According to Google, the summer of 2025 marks an inflection point, calling it the season in which AI became a true game changer for cyber defence.

AI empowers security defenders

“AI provides an unprecedented opportunity for building a new era of American innovation,” says Kent Walker, President of Global Affairs at Google and Alphabet.

“We can use these new tools to grow the US economy, create jobs, accelerate scientific advances and give the advantage back to security defenders.

“When it comes to security opportunities, we’re thrilled to be driving progress in three key areas… agentic capabilities, next-gen security model and platform advances and public-private partnerships focused on putting these tools to work.”

The power of agentic AI

One area Kent and Google feel helping usher in a new era for security is agentic AI — artificial intelligence systems that can autonomously make decisions and take actions.

One of Google’s standout initiatives, Big Sleep, is an AI agent developed by Google DeepMind and Project Zero. Unlike traditional security tools, Big Sleep autonomously scans codebases, actively hunting for unknown “zero-day” software vulnerabilities. 

By the end of last year, it was already finding real-world bugs — plugging security holes before attackers could exploit them 

But what sets Big Sleep apart? 

Its move from detection to pre-emption stands out. 

In partnership with Google Threat Intelligence, this AI agent recently uncovered a critical SQLite vulnerability (CVE-2025-6965) before it could be exploited — cutting off a real-world threat at the source. 

“We believe this is the first time an AI agent has been used to directly foil efforts to exploit a vulnerability in the wild,” Kent says.

Beyond securing Google’s own products, Big Sleep is being deployed to help protect open-source software, extending its benefits across the global technology ecosystem. 

By automating vulnerability discovery, AI liberates human experts to concentrate on the most complex, evasive threats, multiplying the impact of every security team.

Google’s new AI security capabilities

In light of the changing and growing nature of the security landscape, Google is expanding agentic powers to several tools and platforms designed for security professionals.

Google’s new launches:

• Timesketch: Google’s open-source digital forensics platform is being upgraded with AI-driven investigation capabilities. The integration of Sec-Gemini allows the tool to automatically analyse logs and conduct initial forensic investigations, streamlining incident response and reducing the burden on analysts.
• FACADE: Standing for Fast and Accurate Contextual Anomaly Detection Environment, FACADE has been in use at Google since 2018. It leverages AI to sift through billions of security events daily, identifying insider threats without relying on data from past attacks. Its advanced contrastive learning approach means it can spot anomalies without needing examples of known threats, offering a major leap in real-time security monitoring.
• AI-Powered Capture the Flag (CTF) Challenges: At DEF CON 33 in August, Google and Airbus will unveil a CTF event pairing cybersecurity professionals with AI assistants. This is designed to showcase how AI can enhance defenders’ skills across all experience levels, helping to bridge the expertise gap in cybersecurity.

The importance of public-private partnerships

These advancements in AI and cybersecurity further solidify the integral nature of collaboration.

In recognition that no single entity can secure the web alone, Google partnered with industry leaders including Amazon, Anthropic, Cisco, IBM, Intel, Microsoft, NVIDIA, OpenAI and PayPal to launch the Coalition for Secure AI (CoSAI). The coalition sets critical standards for safe AI development and deployment. 

As well as this, Google is also donating data from its Secure AI Framework (SAIF) to accelerate research in agentic AI, cyber defence and software supply chain security.

Kent says: “We have always believed in AI’s potential to make the world safer, but over the last year we have seen real leaps in its capabilities, with new tools redefining what lasting and durable cybersecurity can look like.

“This summer’s advances in AI have the potential to be game-changing, but what we do next matters. 

“By building these tools the right way, applying them in new ways and working together with industry and governments to deploy them at scale, we can usher in a digital future that’s not only more prosperous, but also more secure.”