All news in category “AI and Security Pulse”

🔐 The UK NCSC's chief executive Richard Horne told the RSA Conference (March 24) to 'seize the disruptive vibe coding opportunity' while urgently developing safeguards. He warned AI-assisted development can either reduce systemic vulnerability or propagate new flaws depending on model design and controls. NCSC CTO David C published Secure Vibe Coding Commandments advocating secure-by-default models, provable provenance, AI-powered audits, deterministic guardrails and sandboxed hosting.

🧭 This article presents a practical framework for governing AI agents by aligning user, developer, role-based, and organizational intent. It prescribes a precedence model—organization, role, developer, then user—to resolve conflicts and preserve security and compliance. The authors illustrate expected agent behaviors (refuse, escalate, clarify, or proceed) and advocate for guardrails, least-privilege access, continuous evaluation, telemetry, and human-in-the-loop controls to sustain safe, reliable agent operations.

🔒 Gartner's inaugural Market Guide for Guardian Agents defines a new enterprise control layer that supervises AI agents to keep their actions aligned with organizational goals and boundaries. The article stresses risks from unmanaged non-human identities—so-called identity dark matter—and lists mandatory capabilities across visibility, continuous assurance, and runtime enforcement. It urges enterprises to adopt an enterprise-owned guardian layer rather than relying solely on platform-native controls.

🤖 Japan’s recent election highlighted Team Mirai, a new political party that used AI-driven engagement tools to deepen civic participation and shape policy. Its AI Interviewer walked voters through issues, while the gamified Action Board mobilized volunteers and the open-source Gikai portal summarized legislation. Mirai collected tens of thousands of public inputs and converted them into a manifesto focused on democratic reform rather than traditional left‑right politics.

🤖 Early this year, enterprises began experimenting with autonomous, agentic tools such as Anthropic’s Claude Cowork and the open-source OpenClaw, which can access apps, files and the web to execute multi-step workflows on users’ behalf. Proponents highlight large efficiency gains and the ability to offload routine IT tasks to non-technical staff, while security researchers warn of misalignment, prompt‑injection flaws and unintended destructive actions. IT leaders are advised to permit controlled experimentation, enforce strict permissions and monitoring, and invest in clean operational context to reduce amplified mistakes and limit shadow‑AI risk.

🛡️ AI-driven honeypots pair large language models with deception servers to create dynamic, realistic environments that keep attackers engaged longer and collect richer threat intelligence. Academic research by Dr. M. Abdullah Canbaz and others showed LLMs can parse traffic and handle complex Linux commands, prompting open-source and commercial efforts such as Beelzebub and Deutsche Telekom’s T-Pot. These systems significantly lower the cost and engineering effort of high-interaction deception while enabling deployment in novel locations like APIs and AI agents. However, defenders must balance benefits with risks—attackers are using AI to automate attacks and may develop countermeasures such as deception-detection services or data poisoning—so CISOs should view AI honeypots as a complement to existing sensors and an important tool for improved visibility and hunting.

📚 OpenAI has begun rolling out a new ChatGPT Library feature that stores personal files and images in its cloud so they can be referenced in future chats. The feature is available to Plus, Pro, and Business subscribers worldwide except in the European Economic Area, Switzerland, and the United Kingdom. Files uploaded in chats or via the composer are saved by default to a secure, dedicated location and remain in the Library until manually deleted; deleting a chat does not remove the stored file.

🚨 New ISACA research finds that 56% of IT and cybersecurity professionals cannot say how quickly they could shut down AI systems after a cyber-attack or security incident. The global survey of over 3,400 security and digital professionals found just 32% believe they could halt compromised AI within an hour, and 7% expect it would take longer. Respondents reported confusion over AI ownership, with many unsure who is accountable, limited human oversight of AI actions, and mixed confidence in their organisation's ability to investigate and explain serious AI incidents.

🧠 This article presents a blueprint architecture for securing AI data centers and AI factories as enterprises shift from consuming AI to producing it. It explains how organizations can protect LLMs, data pipelines, and compute infrastructure against emerging, AI-specific threats by combining network segmentation, identity and access controls, data governance, and advanced threat prevention. Check Point emphasizes operational practices and industrial-grade security controls to enable secure, revenue-generating AI deployments.

🔒 At RSAC, Check Point outlined how enterprises can secure their AI transformation by protecting sensitive data, models, and communications from emerging attack vectors. The company emphasized integrated controls—data loss prevention, access governance, model protection and runtime threat detection—paired with unified visibility across cloud and on‑prem environments. Check Point shared practical steps for safe AI adoption, from discovery and policy enforcement to automated response, helping organizations reduce leakage and compliance risk while enabling productive AI use.

🔒 XM Cyber researchers identified eight validated attack vectors inside AWS Bedrock, showing that integrations and permissions — not the foundation models themselves — are the primary risk. The team highlights log manipulation, knowledge base compromise, agent hijacking, flow injection, guardrail degradation, and prompt poisoning as practical paths to data exfiltration and operational abuse. Their findings show how a single over-privileged identity can redirect logs, steal credentials, or subvert agents and prompts. Security teams should inventory AI workloads, enforce least privilege, and map cross-environment attack paths to reduce exposure.

🔒 CrowdStrike announced a suite of innovations that expand AI detection and response across endpoints, SaaS, and cloud environments. New capabilities include runtime monitoring for desktop AI applications and Copilot Studio agents, unified discovery and classification of AI agents across SaaS, and data-flow visibility for cloud-hosted AI workloads. Several features are in pre-beta or early beta with staged GA rollouts planned over upcoming quarters.

🔐At the NRF Big Show in January 2026, Google introduced the Universal Commerce Protocol (UCP) and highlighted compatibility with the Agent Payments Protocol (AP2), promising tokenized payments and verifiable credentials. Unit 42 warns that indirect prompt injection—where agents ingest hidden instructions while browsing—can enable novel fraud such as gift card payload poisoning and refund logic hijacking. Industry forecasts (Bain, McKinsey) predict substantial agentic commerce adoption, increasing the attack surface. Recommended mitigations include protocol guardrails (AP2), Know Your Agent, agent reputation scoring, Unit 42 AI Security Assessments and Prisma AIRS.

🛑 Google will no longer accept AI-generated bug reports for the Open Source Software Vulnerability Reward Program it funds, citing a rising number of low-quality submissions that often contain hallucinated exploit paths or issues with minimal security impact. To reduce triage overhead, some reward tiers will now require higher-quality proof such as an OSS-Fuzz reproduction or a merged patch. Google says this will help teams focus on high-impact, verifiable vulnerabilities. Separately, the company is contributing to programs that use AI constructively to strengthen open-source security.

🔍 Microsoft introduces CTI-REALM, an open-source benchmark that evaluates AI agents on end-to-end detection engineering by turning real-world cyber threat intelligence into validated detections. The benchmark places agents in realistic, tool-rich environments where they must read CTI reports, explore telemetry, iterate on KQL queries, and produce Sigma rules and KQL-based logic scored against ground truth across Linux, AKS, and Azure. CTI-REALM's checkpoint-based scoring surfaces whether failures arise from CTI comprehension, technique mapping, data-source selection, or query construction, helping teams decide where human oversight and guardrails are required.

🛡️ AI-enabled cyber attacks increasingly mimic legitimate users, rendering signature- and rule-based defenses insufficient. Modern identity security must adopt continuous, context-aware risk modeling that evaluates identity, device and session context in real time to detect subtle deviations. Organizations should extend monitoring across cloud, endpoints and privileged accounts, enforce Just-in-Time (JIT) access and consolidate behavioral analytics with session monitoring and granular controls to limit credential abuse and insider misuse.

🔒 Ceros, an AI Trust Layer from Beyond Identity, runs alongside Claude Code on developers' machines to provide real-time visibility, runtime policy enforcement, and cryptographically signed audit records. Installation is non-disruptive—two CLI commands and a brief enrollment tie sessions to verified human identities with hardware-bound keys. The admin console surfaces conversation transcripts, tool invocations, MCP server connections, and signed activity logs that support compliance.

🤖RSA Conference 2026 reframes AI from a single track to the event itself, with roughly 40% of sessions AI-weighted and artificial intelligence woven across identity, cloud, threat intelligence and human-focused tracks. CISOs face a dual mandate: accelerate AI adoption to remain competitive while protecting the enterprise from new attack surfaces such as RAG pipelines, vector databases, prompt injection and model inversion. Key priorities at RSAC include securing the AI stack, defining AI governance and compliance (including preparation for the EU AI Act), managing non‑human identities, mitigating shadow AI and AI-assisted coding risks, and preparing SOCs for autonomous remediation.

🔒 Falcon AIDR now integrates with NVIDIA NeMo Guardrails to provide programmable runtime protections for homegrown AI agents moving into production. The combined solution blocks prompt injection, redacts PII, defangs malicious domains, and moderates unwanted topics while preserving responsive, sub-100ms agent workflows. Teams can leverage 75+ built-in detectors or create custom policies to monitor in report-only mode and then progressively enforce blocks, redactions, encryptions, or transformations.

🔒 Unit 42 examines the security tradeoffs of agentic AI, spotlighting the early 2026 Clawdbot surge and pervasive vulnerabilities such as exposed gateways, plaintext credentials, and overbroad permissions. The piece identifies two primary threat paths: malicious model files and compromised Model Context Protocol (MCP) servers, and explains how compromised agents can act as powerful insider threats. Practical guidance includes scanning and sandboxing models, preferring trusted remote MCPs or auditing local MCP code, enforcing strict least-privilege tool access, implementing prompt-injection guardrails, and maintaining detailed logging and policy reviews.