With the continuous deepening of digital life, cybersecurity not only concerns personal privacy and property safety, but also affects the construction of a clean online environment. As the main provider of digital services, internet platforms bear an irreplaceable and important responsibility in terms of security guarantees. Facing increasingly covert and complex cyber criminal activities and fraud methods, building a multi-party co-governance model based on platform technology and involving wide participation of users has gradually become an industry consensus.
On December 30th, the QQ Security Center released the "2025 Work Report", systematically reviewing QQ's explorations and practices in areas such as the application of underlying large models, account security protection, anti-gang activities management, and user ecosystem collaboration over the past year. Through specific data and typical cases, the report presented the actual achievements brought about by the iterative improvement of technical capabilities and the optimization of governance mechanisms, and demonstrated QQ's positive attitude and efforts in fulfilling platform responsibilities and strengthening the cybersecurity defense line.
The deployment and application of large-scale models will strengthen the security defense line.
Currently, the application of AI technology in the field of cybersecurity is continuously deepening. With its efficient identification capabilities and rapid response advantages, it is widely used in various scenarios such as risk assessment and user communication. In 2025, the QQ Security Center is actively exploring the application of AI large model technology in security governance scenarios. In terms of content governance and anti-gang activities, the platform, based on the underlying large model, conducts debugging and optimization around key security areas such as pornography and fraud, and gradually applies it to actual risk control processes, effectively improving the handling efficiency while reducing the occurrence of misjudgments.
The report indicates that QQ has cumulatively removed over a billion pieces of malicious and vulgar content, as well as over ten thousand pieces of content involving minors' risks throughout the year. In terms of combating illegal industries, the platform has continuously strengthened its prevention capabilities. In 2025, it successfully resisted multiple concentrated attacks by illegal industry groups, removed over one million maliciously hacked accounts, submitted over 300 valid clues to the police, and provided over ten thousand pieces of risk information related to illegal industries to third-party platforms, helping to build a broader security defense mechanism and reducing risks at the source.
In addition, QQ has continued to improve the security reminder mechanism for the login process and enhance the safety education on password management for users. In 2025, QQ successfully blocked and thwarted hundreds of brute-force attack and password guessing attacks. The daily number of stolen accounts on the platform decreased by 30% compared to the previous year.
System capabilities have been upgraded, and the security services have become more personalized.
While strengthening technical defense, the QQ Security Center has also continuously expanded its security capabilities to the user side, and has been strengthening the protection layout in key scenarios such as chat and interaction. Relying on AI and big data capabilities, the platform has built an active prevention mechanism covering pre-event alerts and in-event interventions, helping users identify potential risks.
In response to increasingly covert and scenario-based fraud methods, the platform has systematically analyzed typical fraud links and expanded the coverage scenarios of QQ risk alerts. By 2025, QQ will actively issue over 3 million security alerts daily. After users receive risk warnings, they can also quickly access the "QQ Anti-Fraud Assistant". This is a dialogue AI assistant based on the Hyming large model, which deeply learns the anti-fraud service experience of Tencent customer service over many years. It can provide personalized services to users and help them detect scams. In 2025, the "QQ Anti-Fraud Assistant" has served over 7.7 million users, effectively helping users avoid potential losses in various fraud scenarios such as impersonating acquaintances, false part-time jobs, and emotional inducement.
A user shared that when they received the so-called "time-limited discount coupon", the platform triggered a risk identification and promptly issued a risk reminder. After clicking on "Consult QQ Anti-Fraud Assistant", the anti-fraud assistant further helped the user break down the risk points in the operation process, successfully dissuading the user from continuing the operation.
In terms of protecting minors, the platform has launched 18 different-themed "Flame Care Cards" to address common risks, providing timely and safe reminders and protection for teenagers, and fostering their healthy growth.
Regarding the issue of account recovery that users are concerned about, QQ has continuously optimized the login and verification processes, expanded various reliable verification methods, and assisted users who were unable to log in due to forgotten passwords or changing phone numbers to restore their accounts. One user shared his "account recovery story": The Tencent customer service team helped recover an account that had not been logged in for many years. There were voice messages left by his grandmother before her death, becoming her last cherished memory.
User participation has increased, and a co-governance model is gradually taking shape.
In the face of increasingly complex cybersecurity issues, the joint participation of users is also an important force in safeguarding a clear and clean online environment. The report shows that in 2025, the security awareness of QQ users continued to improve, and their proactive preventive behaviors significantly increased. Throughout the year, tens of millions of users actively reported illegal content. Driven by convenient functions such as "one-click reporting", the willingness of users to report increased by 12% year-on-year, and over 500,000 users reported more than 10 times. The alertness of users to risks has also continuously improved, with nearly 50 million users actively clicking on security alerts throughout the year. At the same time, more than 50,000 users formed "family protection circles", enabling the sharing of risk information among relatives and friends. Once a member within the circle triggers a risk alert, other members will receive the warning synchronously. Through mutual assistance, the scope of security protection is further expanded.
QQ stated in the report that in the future, the platform will continue to increase investment in technological upgrades, promote the in-depth application of cutting-edge technologies such as artificial intelligence and big data in security protection, and continuously optimize the feedback and response processes for users. At the same time, it will continue to strengthen collaboration with users and all sectors of society, enhance safety education and publicity, build a more efficient and effective network governance ecosystem, and work with all parties to jointly create a safer and more trustworthy online environment.
