Intel Centre

Cloudflare Blog reports that cloudflare is shifting bot mitigation from point-in-time Risk assessment to continuous Trust evaluation. Learn how new good and bad behaviors from bots and agents are assessed by our systems, including BotBase and Precursor — and try out our Precursor Trace The leadership question is whether this expands what agents can read, decide, or do without a human in the loop. Review the affected workflow and decide which sources of context should be allowed to influence privileged actions.

arXiv cs.CR agent security reports that tool use transforms LLMs into agents that act beyond their training data, and for code-capable models, programmatic tool calling extends this further by replacing rigid JSON calls with scripts that chain and parallelize naturally. The leadership question is whether this expands what agents can read, decide, or do without a human in the loop. Review the affected workflow and decide which sources of context should be allowed to influence privileged actions.

arXiv cs.CR agent security reports that deciding which of two agents is stronger means playing games until skill outweighs luck, and every game costs money, model inference, or expert time. The leadership question is whether this expands what agents can read, decide, or do without a human in the loop. Review the affected workflow and decide which sources of context should be allowed to influence privileged actions.

arXiv cs.CR agent security reports that lLM-based agentic systems have shown remarkable capabilities in complex domains, while suffering from cascading errors and difficulty in debugging. The leadership question is whether this expands what agents can read, decide, or do without a human in the loop. Review the affected workflow and decide which sources of context should be allowed to influence privileged actions.

arXiv cs.CR agent security reports that large language models (LLMs) increasingly support complex professional tasks, yet their capabilities in rule-intensive document review remain insufficiently evaluated. Multi-agent systems can spread evaluator bias and bad assumptions across the workflow, creating correlated failures instead of independent checks. Avoid treating another model's judgement as neutral evidence. Log provenance and diversify evaluation paths for high-impact decisions.

arXiv cs.CR agent security reports that retrieval-augmented generation over long documents is dominated by one design: chunk the text, embed the chunks, and surface the top-k nearest neighbours of the query. The leadership question is whether this expands what agents can read, decide, or do without a human in the loop. Review the affected workflow and decide which sources of context should be allowed to influence privileged actions.

arXiv cs.CR agent security reports that as LLMs are increasingly deployed within agentic systems, their capabilities depend not only on the model weights but also on the harness: the prompts, tools, control flow, memory, and orchestration code surrounding them. This is the uncomfortable version of agent risk: a normal-looking web page can become the first step in host compromise when the agent has local execution paths. Separate browsing from execution, inspect tool arguments, and replay browser-to-host exploit paths during security review.

Cloudflare Blog reports that the next version of MCP has a rewritten, stateless core that just works on Workers. We cover upgrades to the protocol, the new feature lifecycle and SDK migration path, and hear from early adopters already running it in The leadership question is whether this expands what agents can read, decide, or do without a human in the loop. Review the affected workflow and decide which sources of context should be allowed to influence privileged actions.

Cloudflare Blog reports that more than half of requests now come from machines, not people. Agent Readiness shows how well agents can discover and read your site, while Answer Engine Optimization tracks how often AI assistants recommend The leadership question is whether this expands what agents can read, decide, or do without a human in the loop. Review the affected workflow and decide which sources of context should be allowed to influence privileged actions.

Cloudflare Blog reports that agents are a new kind of visitor. They don't render CSS or click ads, but they have a paying human on the other end. Block them and you block your customer. We're building the open tools and protocols so publishers and agents can cooperate and not The leadership question is whether this expands what agents can read, decide, or do without a human in the loop. Review the affected workflow and decide which sources of context should be allowed to influence privileged actions.

Cloudflare Blog reports that we should be giving all agents tools that excel at what’s important for an AI model. Kitesurf is Cloudflare’s new stateless, highly scalable, and cost-effective web browser that runs entirely on top of Workers and was designed specifically for the Agentic The leadership question is whether this expands what agents can read, decide, or do without a human in the loop. Review the affected workflow and decide which sources of context should be allowed to influence privileged actions.

Cloudflare Blog reports that today we're launching a developer preview of WebMCP on Cloudflare. With one switch, any site becomes usable by browser AI agents — no new APIs, no origin changes — while the human stays in control and creators keep their The leadership question is whether this expands what agents can read, decide, or do without a human in the loop. Review the affected workflow and decide which sources of context should be allowed to influence privileged actions.

arXiv cs.CR agent security reports that long-horizon reasoning requires an agentic runtime that can persist when evidence supports its current approach and pivot when measurements reveal failure, hidden constraints, or a misspecified objective. This is the uncomfortable version of agent risk: a normal-looking web page can become the first step in host compromise when the agent has local execution paths. Separate browsing from execution, inspect tool arguments, and replay browser-to-host exploit paths during security review.

arXiv cs.CR agent security reports that context lengths of language models (LMs) have dramatically increased, driven by the demands for in-context learning, self-improvement, and long-horizon agentic workflows. This is the uncomfortable version of agent risk: a normal-looking web page can become the first step in host compromise when the agent has local execution paths. Separate browsing from execution, inspect tool arguments, and replay browser-to-host exploit paths during security review.

arXiv cs.CR agent security reports that long context reasoning in large language models (LLMs) is usually constrained by the fact that a single inference trajectory has to simultaneously explore the context, store intermediate state, verify evidence, and produce the final answer. The leadership question is whether this expands what agents can read, decide, or do without a human in the loop. Review the affected workflow and decide which sources of context should be allowed to influence privileged actions.

arXiv cs.CR agent security reports that prompt injection poses significant security risks to LLM agents. Efficient and effective red-teaming is therefore critical, both for evaluating these risks and for collecting training data to improve defenses. Existing state-of-the-art prompt injection This is the uncomfortable version of agent risk: a normal-looking web page can become the first step in host compromise when the agent has local execution paths. Separate browsing from execution, inspect tool arguments, and replay browser-to-host exploit paths during security review.

arXiv cs.CR agent security reports that aI-supported care planning can help clinicians, patients, caregivers, and care teams coordinate complex decisions across clinical, functional, psychosocial, and environmental needs. Multi-agent systems can spread evaluator bias and bad assumptions across the workflow, creating correlated failures instead of independent checks. Avoid treating another model's judgement as neutral evidence. Log provenance and diversify evaluation paths for high-impact decisions.

arXiv cs.CR agent security reports that agents for long term reasoning require a memory that can be efficiently and effectively updated over time, as new facts and external feedback continue to arrive. The leadership question is whether this expands what agents can read, decide, or do without a human in the loop. Review the affected workflow and decide which sources of context should be allowed to influence privileged actions.

arXiv cs.CR agent security reports that sparse matrix kernels are fundamental to scientific computing, graph analytics, and machine learning. Their GPU performance depends strongly on the input sparsity pattern and execution strategy. For the same SpMM on the same matrix, cuSPARSE exhibits a 350x This is conventional infrastructure risk, but it matters to agent teams when agents can trigger workloads on affected hosts. Patch the affected estate and keep agent-run execution environments isolated from sensitive infrastructure.

arXiv cs.CR agent security reports that high-quality annotation of artworks is essential for computational art research, yet extracting implicit semantics remains challenging due to the reliance on culturally grounded meanings and deep contextual knowledge behind the images. Multi-agent systems can spread evaluator bias and bad assumptions across the workflow, creating correlated failures instead of independent checks. Avoid treating another model's judgement as neutral evidence. Log provenance and diversify evaluation paths for high-impact decisions.

arXiv cs.CR agent security reports that the capabilities of an LLM agent depend not only on its model but on the harness: the executable program that constructs context, invokes tools, verifies results, and recovers from failure. The leadership question is whether this expands what agents can read, decide, or do without a human in the loop. Review the affected workflow and decide which sources of context should be allowed to influence privileged actions.

Cloudflare Blog reports that at Cloudflare, we knew we could not depend on every employee to configure every agent perfectly or watch every tool call. Before expanding write access across our own internal MCP servers, we built WriteGuard. We are now bringing those controls to Cloudflare The leadership question is whether this expands what agents can read, decide, or do without a human in the loop. Review the affected workflow and decide which sources of context should be allowed to influence privileged actions.

arXiv cs.CR agent security reports that recursive self-improvement requires agents to turn accumulated experience into better future behavior. Personal AI agents offer a concrete setting for studying this capability because they retain preferences, task histories, tool routines, and learned skills The leadership question is whether this expands what agents can read, decide, or do without a human in the loop. Review the affected workflow and decide which sources of context should be allowed to influence privileged actions.

arXiv cs.CR agent security reports that human input reaches language models by typing or speaking, and each channel leaves a distinct signature: orthographic noise for keyboards; for voice, disfluency from conventional transcription and restructuring from AI-backed dictation tools. The leadership question is whether this expands what agents can read, decide, or do without a human in the loop. Review the affected workflow and decide which sources of context should be allowed to influence privileged actions.

arXiv cs.CR agent security reports that as autonomous agents powered by foundation models are increasingly integrated into social and economic systems, understanding the principles governing their collective behavior is essential for ensuring safety and cooperation. The leadership question is whether this expands what agents can read, decide, or do without a human in the loop. Review the affected workflow and decide which sources of context should be allowed to influence privileged actions.

Microsoft Security Blog reports that microsoft expands its Zero Trust for AI strategy to enhance security for AI and DevSecOps environments with new tools and guidance. The post Advance Zero Trust for AI: New tools and guidance to secure AI agents and DevSecOps appeared first on Microsoft The leadership question is whether this expands what agents can read, decide, or do without a human in the loop. Review the affected workflow and decide which sources of context should be allowed to influence privileged actions.

Cloudflare Blog reports that cloudflare Wallets will provide AI agents with native payments and verifiable identity on the web. Using the x402 protocol, agents can autonomously purchase APIs and content within clear safety The leadership question is whether this expands what agents can read, decide, or do without a human in the loop. Review the affected workflow and decide which sources of context should be allowed to influence privileged actions.

Cloudflare Blog reports that learn how to build customizable, sandboxed CI/CD pipelines natively on Cloudflare using Workflows, Artifacts, and the CI SDK. We walk through replacing complex YAML configurations with TypeScript workflow steps and self-healing AI The leadership question is whether this expands what agents can read, decide, or do without a human in the loop. Review the affected workflow and decide which sources of context should be allowed to influence privileged actions.

Cloudflare Blog reports that we created the Cloudflare Codex, a governed body of engineering standards that AI agents consume across the development lifecycle. By pairing structured RFCs with agentic reviews, teams automatically enforce consistency across code, specs, and incident This is the uncomfortable version of agent risk: a normal-looking web page can become the first step in host compromise when the agent has local execution paths. Separate browsing from execution, inspect tool arguments, and replay browser-to-host exploit paths during security review.

arXiv cs.CR agent security reports that in [AS21], Axiotis and Sviridenko conjectured that the linear dependence on the restricted condition number in sparse convex optimization cannot be improved by a polynomial-time algorithm. The leadership question is whether this expands what agents can read, decide, or do without a human in the loop. Review the affected workflow and decide which sources of context should be allowed to influence privileged actions.