AI Security Transformation Driven by Computing Power: In-depth Analysis of the AI Industry Market and Development Strategy Research
-
The management changes at OpenAI this weekend have attracted significant attention. The latest news indicates that OpenAI is 'optimistic' about Sam Altman and Greg Brockman returning to the company, with the interim CEO extending goodwill toward Altman.
Analysts suggest that the root of this internal conflict may lie in the tension between AI regulation and commercialization. OpenAI's Chief Scientist, Ilya Sutskever, reportedly could not accept CEO Sam Altman's aggressive commercialization strategy, leading him and other board members to oust Altman and company president Greg Brockman.
CITIC Securities notes that Ilya has shifted his focus from developing the next-generation GPT/DALL·E models to preventing AI models from going rogue. This may imply that OpenAI's investments in GPT-5 and DALL·E 4 are being partially redirected toward AI safety and alignment efforts.
TF Securities also analyzed that this management shift at OpenAI marks a pivotal event in the era of large AI models. It signifies that OpenAI, as an industry leader, is accelerating its model capabilities while placing greater emphasis on safety and accessibility. However, commercialization efforts may increasingly be delegated to industry partners.
Additionally, as of November 18, the third-party website gptstore.ai has collected 12,967 shared GPTs, showing explosive growth. OpenAI's new services have exceeded expectations in popularity, but combined with DDoS attacks, ChatGPT servers experienced intermittent widespread outages from November 6 to 8 (Pacific Time).
Guosheng Securities points out that this is not the first time OpenAI has been constrained by computing power bottlenecks. From early 2023 to October, ChatGPT experienced six outages, and its API suffered seven outages. Computing power is the foundational resource for AIGC development and remains a critical bottleneck in training and scaling. Chip manufacturers, server providers, computing power leasing services, and optical module suppliers are expected to benefit significantly. Meanwhile, the diffusion of technology will also benefit AI applications in the long run.
On the corporate front, according to public company disclosures:
- NSFOCUS has initiated research on ChatGPT-generated content detection tools and is participating in the formulation of AI technology standards.
- Anbotong has officially partnered with Baidu Security for AI security services.
- Kingsoft Office's WPS AI has entered public testing.
Facial recognition, voice recognition, intelligent algorithm recommendations—AI technologies are rapidly integrating into daily life. At the beginning of 2023, ChatGPT garnered widespread attention for its vast knowledge and exceptional natural language interaction capabilities.
With the booming development of the AI industry, massive amounts of data are being collected, stored, and utilized in people's lives. Consequently, data security and privacy protection have become critical challenges in the development and application of AI systems.
China Telecom, Ant Group, Huawei, Baidu, ByteDance and Over 30 Institutions Jointly Initiate AI Security Working Group
At the 'Exploring Next-Generation Security Intelligence' forum held during the 2023 inclusion·Bund Conference, the Cloud Security Alliance (CSA) Greater China Region announced the establishment of the 'AI Security Working Group.' Over 30 institutions, including China Telecom, Ant Group, Huawei, Baidu, ByteDance, Xidian University, and the National Financial Evaluation Center, became the first batch of initiating members.
According to the CSA official website, the main responsibilities of the working group include researching and compiling materials related to AI security threats and risks, AI data security and privacy protection, AI technology applications for security, AI ethics, and AI compliance. The working group will focus on various AI technology threats, risks, protections, and other related issues centered around AI security.
The application of AI technology has brought many conveniences but has also introduced a series of security risks. The establishment of the AI Security Working Group aims to strengthen research and protection of AI technology security, ensuring the safety and privacy of people using AI technology.
Currently, a new round of technological revolution and industrial transformation is emerging. Driven by technologies such as 5G, big data, cloud computing, and deep learning, artificial intelligence (AI), as a critical strategic technology for new infrastructure, is accelerating its development and integrating innovatively with various industries, triggering chain-like transformations.
Particularly in the field of cyberspace security protection, AI has unique value and advantages in threat identification, situational awareness, risk scoring, malicious detection, harmful information governance, spam and fraud call detection, and gray/black industry identification. The demand for AI applications in these areas is experiencing leapfrog development, generating significant spillover effects.
AI Technology Deepens and Accelerates Security Transformation
From the perspective of product layout and application fields, the main competitors in China's AI security industry have similar product layouts, primarily focusing on big data analytics-driven security management platform products. From a technical perspective, Chinese companies' products are based on big data platforms, collecting diverse and heterogeneous massive logs and utilizing technologies such as correlation analysis, machine learning, and threat intelligence to provide security operation tools for security personnel.
Industry enterprises rely on integrated modeling scenarios such as correlation, rules, and statistics for complete conception, mastery, and implementation. Through active monitoring, precise protection, intelligent analysis, and operational collaboration, they achieve proactive management of security objects and timely analysis of security events, with little difference from major domestic competitors.
This year, the highly popular large models have become a fiercely competitive arena, showcasing a 'hundred-model battle' scenario.
Recently, NSFOCUS unveiled its 'Fengyun Guardian' security large model and the Technical White Paper on Security Industry Large Model SecLLM. This vertical security model is trained on vast amounts of security expertise, forming a cybersecurity operations decision-support system that covers scenarios such as security operations, detection and response, offensive and defensive对抗, and knowledge Q&A.
Liu Wenmao, General Manager of NSFOCUS Innovation Research Institute, explained that the three enhanced capabilities of general large language models (LLMs)—general knowledge semantics, logical analysis, and interactive decision-making—play a critical role in the security field. The cybersecurity industry requires speed, comprehensive coverage, and professional depth to address diverse customer needs in major保障, security operations, and emergency response scenarios.
This large model demonstrates six major advantages: enhanced high-quality security datasets, efficient parallel acceleration for incremental training, efficient fine-tuning for downstream tasks, low-computing-power inference acceleration, human-machine alignment for security control, and secure private-domain interactions.
In terms of security intelligence, NSFOCUS released the 'Fengyun Guardian' security large model, trained on extensive security专业知识, to build a decision-support system for cybersecurity operations across multiple scenarios.
Additionally, the model is developed across three dimensions: 'scenario牵引, technology-driven, and ecosystem synergy.' It leverages AI to reshape the security industry's workflow, addressing three key challenges through large models:实战态势 command and dispatch, red-blue对抗辅助决策, and security operations efficiency improvement.
China's cybersecurity industry has grown significantly over the years, offering companies opportunities to expand their market share. However, as user demands for cybersecurity products and services increase, competition intensifies due to the growing scale and competitiveness of existing players and the entry of new competitors.
AI security industry research reports primarily analyze the market size, supply-demand dynamics, competitive landscape, and key players' market shares, while also providing scientific forecasts for the industry's future development.