How AI is Revolutionizing Security Operations: Insights from the 2025 Global Digital Economy Conference

On July 2, 2025, the Global Digital Economy Conference and Big Data Innovation Application Forum were held at the National Convention Center in Beijing, with the theme “Deeply Mining Big Data Value, Inspiring New Momentum for the Digital Economy”. The event focused on the integration of big data and artificial intelligence to propose new paradigms for high‑quality digital‑economy development.

360 Digital Security Group’s Vice President Yu Kai highlighted that while generative AI models have greatly enhanced security products and operational efficiency, they also introduce unprecedented challenges such as model‑environment vulnerabilities, malicious content generation, hallucinations, and uncontrolled processes. He emphasized that the more AI is used in security, the more urgent the need for robust AI security becomes.

360 is embedding large‑model capabilities into its native security platform, creating an “AI‑native security operation platform” that fuses high‑quality platforms with top‑tier models to reshape network security, data security, and model‑security scenarios. By building a matrix of intelligent risk‑control models—including security assessment, risk detection, and red‑blue‑team confrontation models—360 ensures a closed‑loop lifecycle for data collection, model training, deployment, and iterative updates, keeping large‑model applications compliant and beneficial.

In practice, 360 has launched over 100 security‑digital‑expert agents that automate threat hunting, deep analysis, attack tracing, and phishing detection. These agents have achieved up to 99% noise‑reduction in alert handling, a ten‑fold increase in data‑governance efficiency, and a fifteen‑fold boost in security‑event analysis speed, cutting average investigation time from 120 minutes to 5 minutes and raising automated response rates from 20% to 60%.

Additionally, a provincial government deployment of a large‑model content‑security system covering smart‑government services (Q&A, intelligent writing, digital employees, query and客服) detected and filtered thousands of harmful items daily, achieving a 99% safe‑response rate and a 99% risk‑identification accuracy, thereby safeguarding the baseline of governmental operations and accelerating digital transformation.