Cloudflare’s APAC leadership described an inflection point: AI-agent traffic grew more than 1,700% year over year and was said to account for about 60% of Cloudflare-network traffic—evidence, in its view, that the internet is moving from predominantly human-driven activity to autonomous machine acti Cloudflare’s APA...
Research answer

Create a landscape editorial hero image for this Studio Global article: What did Cloudflare’s Asia Pacific leadership reveal about the 1,700% year over year surge in AI agent traffic to roughly 60% of the company. Article summary: Cloudflare’s APAC leadership described an inflection point: AI agent traffic grew more than 1,700% year over year and was said to account for about 60% of Cloudflare network traffic—evidence, in its view, that the intern. Topic tags: general web, agents, ai, automation, productivity. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks, c
Cloudflare’s APAC leadership described an inflection point: AI-agent traffic grew more than 1,700% year over year and was said to account for about 60% of Cloudflare-network traffic—evidence, in its view, that the internet is moving from predominantly human-driven activity to autonomous machine activity. 9
New risk model: Agents do more than generate text: they can browse, call tools, access corporate systems, and act on instructions. That expands exposure to prompt injection, unintended or excessive actions, credential misuse, and data exfiltration. “Shadow AI”—staff using unapproved AI services—can send proprietary or personal data outside approved controls. Unauthorized “shadow MCP” connections add ungoverned agent-to-tool/data paths, potentially bypassing identity, access, logging, and security review.
Cloudflare’s recommended controls: First, discover AI and MCP use on managed networks rather than assuming it is sanctioned. Cloudflare says Gateway can identify MCP traffic with protocol-level heuristics, enabling security teams to find shadow MCP usage. 11 Second, apply prompt/DLP-style policies to detect and block sensitive information before it is sent to AI services. Third, make approved MCP servers reachable only through a controlled portal and require identity-aware, least-privilege, zero-trust access; block direct connections to those servers from managed paths. 11
Quantum timeline: “Q Day” means the point at which cryptographically relevant quantum computers can break widely used public-key cryptography, including the “harvest now, decrypt later” threat to data captured today. Cloudflare’s product-wide target is to be fully post-quantum secure by 2029. 34 It has also begun deploying post-quantum authentication for origin connections through Authenticated Origin Pulls and Custom Origin Trust Store—described as a first step toward PQ authentication across its products. 15 The APAC discussion reportedly framed the risk as potentially arriving around 2030, so the operational message is to migrate before a precise Q-Day date is known. 9
How Kitesurf fits: Kitesurf is Cloudflare’s free-beta, agent-first browser for Browser Run: a stateless browser running entirely on Workers, designed for programmatic agent tasks rather than human visual browsing. 12 By eliminating conventional browser overhead and focusing on functions such as screenshots and HTML extraction, Cloudflare says it uses 3–7× less CPU and memory than Chromium for common agentic workloads. 2 Its use of Workers/V8-isolate infrastructure makes browser automation scalable and closer to Cloudflare’s security, identity, network-policy, and observability controls.
Taken together, the strategy is not merely to defend against AI traffic: Cloudflare is building infrastructure for agents to operate the web efficiently while applying discovery, data controls, zero-trust authorization, and cryptographic migration to the new machine-operated layer of the internet.
Studio Global AI
This page includes a source-backed answer you can continue inside Studio Global.
Cloudflare’s APAC leadership described an inflection point: AI-agent traffic grew more than 1,700% year over year and was said to account for about 60% of Cloudflare-network traffic—evidence, in its view, that the internet is moving from predominantly human-driven activity to autonomous machine acti
Cloudflare’s APAC leadership described an inflection point: AI-agent traffic grew more than 1,700% year over year and was said to account for about 60% of Cloudflare-network traffic—evidence, in its view, that the internet is moving from predominantly human-driven activity to autonomous machine acti Cloudflare’s APAC leadership described an inflection point: AI-agent traffic grew more than 1,700% year over year and was said to account for about 60% of Cloudflare-network traffic—evidence, in its view, that the internet is moving from predominantly human-driven activity to aut
**New risk model:** Agents do more than generate text: they can browse, call tools, access corporate systems, and act on instructions. That expands exposure to prompt injection, unintended or excessive actions, credential misuse, and data exfiltration. “Shadow AI”—staff using una