Zhou Hongyi says cybersecurity underpins national security, and as everything becomes software-based and intelligent, vulnerabilities will inevitably be exploited and attacked.
There is no national security without cybersecurity. Cybersecurity is closely tied to every individual and to the development of every country. Especially once everything becomes software-based and intelligent, there will certainly be vulnerabilities, and vulnerabilities will be exploited and attacked.
Original (zh)
“没有网络安全，就没有国家安全。”全国政协委员、360集团创始人周鸿祎在接受“国家安全大家谈”采访时表示，网络安全跟我们每个人、跟每个国家的发展休戚相关。特别是当所有东西都软件化、智能化之后，一定会有漏洞，有漏洞就会被人利用，就一定会被人攻击。
Zhou Hongyi says future project teams may consist of one leader plus a group of AI employees with different capabilities, with the leader responsible for goals, trade-offs and results.
未来项目组可能是一个负责人加一群不同能力的AI员工，负责人负责目标、取舍和结果。
Zhou Hongyi warns that long-half-life foundational principles and judgment must not be outsourced to AI, or people will be unable to spot AI's mistakes.
长半衰期的基础原理和判断力不能外包给AI，否则无法识别AI的错误。
Zhou Hongyi suggests a company's core competitiveness may no longer depend on headcount but on how many AIs take part in real work internally.
公司核心竞争力可能不再只看员工数量，而看内部有多少AI在参与真实工作。
Zhou Hongyi says enterprises are shifting from training to inference applications, which will drive hundred-fold short-term growth in inference compute that surpasses training compute.
Enterprises no longer frequently train large models from scratch, but instead call AI directly through inference applications to solve real problems; this shift will drive a short-term hundred-fold growth in inference compute demand, exceeding training compute in both scale and growth rate.
Original (zh)
企业不再频繁从头训练大模型，而是更多通过推理应用直接调用AI解决实际问题，这一转变会推动推理算力需求呈现短期百倍级增长，规模和增速都将超越训练算力。
Zhou predicts the AI chip market will split into dual tracks, with international vendors dominating training chips and many vendors competing in cost-advantaged inference ASICs.
Zhou Hongyi judges that the AI chip market will shift from a single-pole to a dual-track structure: high-end training chips remain dominated by international vendors, while multiple vendors will participate in inference chips. Dedicated inference chips (ASICs) will quickly penetrate niche scenarios thanks to cost advantages.
Original (zh)
周鸿祎判断，AI芯片市场会从单极格局转向双轨结构：高端训练芯片仍由国际厂商主导，而推理芯片领域会出现多厂商共同参与的局面。专用推理芯片（ASIC）凭借成本优势，会在各类细分场景快速渗透。
Zhou says the looser chip requirements of inference give China's domestic chip industry a chance at differentiated development.
Inference compute places relatively looser requirements on chips, with lower power consumption and more controllable cost, which gives China's domestic chip industry an opportunity for differentiated development.
Original (zh)
推理算力对芯片的要求相对宽松，功耗更低、成本更可控，这为国内芯片产业提供了差异化发展的机会。
Zhou Hongyi says agents are becoming the main vehicle for AI deployment and that inference compute demand is exploding, surpassing training.
We made two forward-looking predictions: first, agents become the main vehicle for AI deployment, with competition shifting from parameter comparison to real-world deployment capability. Second, inference compute demand explodes, with inference usage and growth surpassing the compute needed for model training.
Original (zh)
一是智能体成为AI落地主角，智能体应用走进千行百业，深入生产和业务系统，行业竞争焦点从参数比拼转向落地能力比拼。二是推理算力需求爆发，人们使用智能体完成复杂行业任务或需求时，需要多次调用大模型进行推理，算力消耗巨大，推理算力使用规模、增速全面超越大模型训练所需算力。
Zhou says an agent system let a 10-person team do in three weeks what a 100-person team needed half a year for.
A 10-person team used the rule-abiding, deterministic agent system we built to develop in three weeks a product that used to take a 100-person team half a year, with coding, testing and delivery fully automated.
Original (zh)
使用我们打造的守规矩、守纪律、能保持确定性的智能体体系，一个10人小队用三周时间开发了过去100人团队需半年才能生产的产品，并实现了编码、测试、交付流程的全自动化。
Zhou says programmers won't be replaced but will shift to specifying requirements and standards for AI.
We used to think AI would eliminate programmers. In practice, programmers remain important; their focus shifts from repetitive coding and testing to giving AI requirements and setting standards for AI output. Those who can transition will grow into 'super individuals'.
Original (zh)
过去我们认为AI会淘汰程序员。实践后，我们发现程序员依然重要，其主要精力从写代码和做测试这类重复性工作中，转而从事两类新工作：一是给AI提要求；二是给AI的成果定标准。能转型的程序员不仅不会被淘汰，还会在与AI协作过程中成长为“超级个体”。
Zhou warns that AI agents make vulnerability discovery cheap and automated, upending the offense-defense balance.
AI has reshaped the rules of cybersecurity. Vulnerabilities used to be hard to find, so attacks were infrequent and offense and defense were roughly balanced. Now agents can find vulnerabilities in bulk, automatically, cheaply and quickly; without agent infrastructure of equal capability, information systems will be vulnerable everywhere.
Original (zh)
AI已重塑网络安全的游戏规则。过去漏洞很难被发现，因此攻击频率较低，攻防能勉强维持平衡。但现在智能体可以批量、自动化、低成本、快速发现漏洞，如果没有具备对等能力的智能体基础设施，信息系统将处处是漏洞、处处可攻击，面临更大威胁。
Zhou advocates 'using models to govern models', with security agents monitoring other agents while humans retain final say.
The smarter agents are, the less controllable and more uncertain they become. We need 'using models to govern models': build another security model and security agent to analyze agent behavior and inference logs and predict loss of control, reporting to humans who make the final decision.
Original (zh)
智能体越聪明，越不可控、越具有不确定性，它会产生自己的“想法”，此时就需要“以模治模”，即搜集智能体的行为日志、推理日志，打造另一个安全大模型和安全智能体对此进行分析、判断，对它可能的失控行为预判。
Zhou argues open source is the best way to address AI safety, rather than locking models away.
I believe open source is the best means of solving safety problems. When technology is in the hands of a few, and they refuse to address safety, it creates hidden dangers. Guarding against agent risks is not about 'locking up' models.
Original (zh)
我认为开源是解决安全问题的最好手段。当技术掌握在少数人手中时，如果这部分人拒绝解决安全问题，会带来隐患。防范智能体带来的社会风险，不是把模型“锁”起来，而要强调以人为本、智能向善，安全与普惠并重。
Zhou suggests US AI giants invoke safety to preserve their lead and block China's open-source catch-up.
Some US AI giants with a first-mover advantage have repeatedly cited safety to demand halting AI development. There may be two reasons: they hold core technology and don't want others to have it, and China's open-source AI brings technological inclusiveness that could enable it to overtake the US.
Original (zh)
一些美国AI巨头掌握先发优势，多次以安全为由，要求停止AI开发，这背后可能主要有两个原因：一是他们掌握了核心技术，不希望别人也掌握此类技术；二是中国开源人工智能的发展带来了技术普惠，有可能推动实现对美反超，这是他们不希望看到的。
Zhou Hongyi revealed 360 found permission and execution risks in the OpenClaw architecture and switched to architectures inspired by Codex and Claude Code, stressing agent security.
周鸿祎透露，团队早期曾尝试OpenClaw架构，但在内部使用过程中发现了权限和执行方面的风险，后来调整为借鉴Codex、Claude Code等产品的架构。
Zhou Hongyi has repeatedly said publicly that entrepreneurs building a personal IP is the cheapest form of advertising.
他多次公开讲，企业家做IP是最便宜的广告。
Zhou Hongyi has acknowledged that becoming an internet celebrity means accepting amplified scrutiny, where a single misstatement can be very costly.
周鸿祎自己也说过，做网红要接受被放大审视，说错一句话代价很大。
Zhou Hongyi has said repeatedly in short videos that the security industry will be pushed out if it does not use AI.
‘安全这个行当，不用AI就出局。’
Zhou Hongyi says the security industry will be eliminated if it does not adopt AI, as attackers can now generate phishing emails and scam voice calls at near-zero cost.
周鸿祎指出，安全行业若不借助AI，将会被淘汰。
Zhou Hongyi says that as open-source models and inference tech mature, enterprises need not train large models from scratch and can deploy AI on their own servers, building agents that understand their business and keeping competitiveness in their own hands.
随着开源模型和推理技术不断成熟，企业不必从零训练大模型，也有机会在自己的服务器上部署AI，打造真正懂自己业务的智能体，把竞争力掌握在自己手里。
Zhou Hongyi disclosed that 360 has built a sovereign AI operating system that is already deployed in some government and enterprise units.
360 has built a sovereign AI operating system, which is already in use at some government and enterprise organizations.
Original (zh)
三六零（601360）已打造一套主权AI操作系统，已在部分政企单位投用
Zhou Hongyi said DeepSeek's full open-sourcing of its DeepInfer inference framework reflects technological confidence and an inclusive open-source philosophy, and at a deeper level signals technological sovereignty in the AI era.
This is a kind of technological confidence, and even more a philosophy of inclusive open source.
Original (zh)
是一种技术自信，更是一种开源普惠的理念
Zhou Hongyi revealed that 360 has built a sovereign AI operating system that is already deployed at some government and enterprise units.
360 has built a sovereign AI operating system, which is already in use at some government and enterprise organizations.
Original (zh)
三六零已打造一套主权AI操作系统，已在部分政企单位投用。
Zhou Hongyi said DeepSeek's open-sourcing of its inference framework signals that the fight for AI technological sovereignty has arrived, and that 360 has built a sovereign AI operating system already used by some government and enterprise units.
Zhou Hongyi also mentioned that looking deeper, this indicates that the struggle for technological sovereignty in the AI era has arrived. He revealed that 360 has developed a sovereign AI operating system, which is currently in use in some government and enterprise units.
Zhou Hongyi says DeepSeek fully open-sourcing its DeepInfer inference framework shows technological confidence and an open-source inclusive philosophy, and signals that the AI-era contest over technological sovereignty has arrived.
This reflects technological confidence, and even more a philosophy of open-source inclusiveness. Looking deeper, it shows that the contest over technological sovereignty in the AI era has arrived.
Original (zh)
针对 DeepSeek 将内部使用的 AI 推理框架 DeepInfer 全面开源，三六零创始人周鸿祎表示，这背后是一种技术自信，更是一种开源普惠的理念。 周鸿祎同时表示，往深一层看，这表明 AI 时代的技术主权之争已经到来。
Zhou Hongyi revealed that 360 has built a sovereign AI operating system, already in use at some government and enterprise organizations.
他透露，三六零已打造一套主权 AI 操作系统，目前已在部分政企单位投入使用。
Zhou Hongyi says the focus of AI competition is moving from model parameter scale to practical, real-world deployment.
The focus of AI competition is shifting from comparing parameters to comparing real-world deployment.
Original (zh)
AI竞争焦点正从比拼参数转向比拼落地
Zhou Hongyi predicts that a small team plus AI agents could do work that once required dozens of people.
In the future, perhaps one boss with a few employees, plus AI agents, will be able to do work that used to require dozens of people.
Original (zh)
未来或许一个老板带着几名员工，再加上AI智能体，就能完成过去几十个人才能做的工作。
Zhou Hongyi says Laoganma's AI visual inspection turns AI into a tireless quality inspector that keeps product quality stable, and restaurants are applying the same idea to kitchen compliance.
Laoganma's AI visual inspection system has made AI a 'tireless quality inspector', ultimately ensuring consistent quality in every bottle of chili sauce.
Original (zh)
周鸿祎表示老干妈上线的AI视觉质检系统让AI成为“不知疲倦的质检员”，最终保障了每瓶辣椒酱的品质稳定性。
Zhou Hongyi argues that agents can act on computers and files, so errors can hit real business, making AI security the key differentiator.
Agents differ from chatbots in that they not only answer questions but can also operate computers, call programs and modify files; once something goes wrong, real business is affected.
Original (zh)
Agent与聊天机器人最大的区别，是前者不只回答问题，还能操作电脑、调用程序、修改文件。一旦出错，影响的可能是企业真正的业务。
Zhou Hongyi admits 360 has not completed its own enterprise AI transformation, having only reached the second of three stages (agents boosting individual efficiency).
But we've only explored to the second step so far, and haven't finished exploring it yet.
Original (zh)
不过我们现在也才探索到中间第二步，还没探索完呢。
Zhou says 360 lacks the strength and talent density to compete on foundation models, so it will focus on enterprise agents and AI security.
Once you have an objective understanding of yourself, you need to find your own battlefield.
Original (zh)
有了客观认知之后，你就要找到自己的战场。
Zhou Hongyi reflected on how, despite long climbing experience, he fell and badly injured his knee, suggesting that even a low, padded fall carries real risk.
“攀岩攀了这么久，怎么就在阴沟里翻船了呢？”
Zhou Hongyi said 360 got burned once investing in new energy vehicles, will not enter that investment area again, and sees larger opportunities in AI.
9月15日至16日，在湖南株洲举办的第五届北斗规模应用国际峰会期间，他接受媒体采访时表示，已经在新能源汽车投资上吃过一次亏，不会再进入这个投资领域，同时认为人工智能具有更大的发展机会。
Zhou Hongyi warns that AI tools like Mythos enable industrial-scale vulnerability discovery in hours, lowering attack costs and possibly triggering a vulnerability explosion.
He stressed that vulnerabilities are the lifeline of cyber offense and defense; with the emergence of AI tools such as Mythos, vulnerabilities can be 'industrially produced', the cycle compressed from years or months to hours, discovery costs lowered, attacks becoming democratized, and the entire network may face a vulnerability explosion.
Original (zh)
他强调，漏洞是网络攻防的命门，随着Mythos等AI工具出现，漏洞可被“工业化生产”，周期从数年数月压缩至数小时，挖掘成本降低，攻击趋平民化，全网或将迎来漏洞大爆发。
Zhou Hongyi warns that hacker skills distilled into AI agents can be mass-replicated, putting China's security industry at risk of a 'second one-way transparency'.
Even more worrying, hacker expertise can be distilled into large models; mass-replicated 'hacker agents' far exceed humans in attack breadth and depth, and China's cybersecurity industry faces a 'second one-way transparency' threat where the adversary is fast and we are slow.
Original (zh)
更需警惕的是，黑客经验可被蒸馏进大模型，被批量复制的"黑客智能体"，攻击广度和深度远超人类，中国网络安全行业面临敌快我慢的“第二次单向透明”威胁。
Zhou Hongyi says AI agents with execution permissions pose loss-of-control risks and new enterprise attack surfaces, amplified by multi-agent collaboration.
Meanwhile, agents hold planning, invocation and execution permissions; once out of control they are themselves a new risk, they create new attack surfaces once inside enterprises, and multi-agent collaboration will further expand risk exposure.
Original (zh)
同时，智能体掌握规划、调用、执行权限，一旦失控本身就是新风险，其进入企业内部后也会形成新的攻击面，多智能体协作将进一步扩大风险敞口。
Zhou Hongyi urges China's security industry to rapidly build indigenous AI vulnerability discovery, automated defense and agent security capabilities.
China's cybersecurity industry must quickly build its own AI vulnerability discovery, automated defense and agent security systems.
Original (zh)
中国网络安全行业必须尽快形成自己的AI漏洞发现、自动化防御和智能体安全体系。
Zhou Hongyi argues AI-era cyber risks span the supply chain and require industry-wide collaboration, as with 360's 'Panshi Shield' plan.
Cyberspace security risks in the AI era span multiple upstream and downstream links, and no single company can tackle them alone.
Original (zh)
AI时代网络空间安全风险跨越上下游多个环节，不可能靠一家企业单打独斗。