Agentic

• ICML 2026 begins July 6 in Seoul, featuring a record-breaking 23,918 submissions, which is double the volume of the previous year. • A peer-review crisis emerged after watermark-based LLM detection identified 398 reviewers violating policies, resulting in 497 desk-rejections. • Agentic AI has become the dominant theme for workshop proposals, reflecting a major shift in the structural direction of AI development.

• Attackers leveraged a vulnerability in Langflow (CVE-2025-3248) to execute a sophisticated ransomware attack powered by agentic AI. • The breach utilized LLM agents to automate a complex, multi-stage intrusion process involving real-time reasoning, reconnaissance, credential theft, and lateral movement. • This incident is significant because it demonstrates how AI agents can combine known exploitation techniques to automate high-level attack chains that previously required manual human effort.

• A Commvault study of 1,234 organizations across eight Asian markets, including Singapore and Hong Kong, reveals a surge in AI investment heading into 2026. • More than one-third of surveyed firms are already trialling or deploying agentic AI to optimize IT, cybersecurity, and general business operations. • Despite this technological push, the report highlights a critical gap in data resilience, noting that governance, identity resilience, and recovery planning are lagging.

• Accenture and ServiceNow have launched a joint offering designed to accelerate the transition from legacy risk platforms to "Agentic AI." • The partnership features two primary components: managed security services powered by the ServiceNow AI Platform and an AI-driven solution to automate migration from legacy systems. • This collaboration matters as it aims to streamline business reinvention by replacing outdated infrastructure with automated, AI-powered workflows.

• Bitdefender's 2026 Cybersecurity Assessment Report, surveying over 1,200 professionals across six countries, identifies internal AI systems, LLMs, and cloud infrastructure as the top security concerns for 45% and 44% of respondents, respectively. • Over the last year, 41.8% of organizations suffered cloud breaches, while 59.2% confirmed AI-driven social engineering attacks; U.S. firms were particularly hit by business email compromise at a rate of 54.7%. • The report highlights critical emerging threats, including self-mutating malware (55.9%), sensitive data leaks into public LLMs (53.5%), and AI-driven evasion of traditional endpoint detection and response (EDR) tools.

• Enterprises are facing escalating costs and governance gaps as thousands of agentic AI agents are deployed across multiple corporate systems to work alongside human staff. • To address these challenges, Kong is developing a new system that matches specific AI workloads with the most appropriate models based on the complexity of the task. • This initiative aims to help organizations optimize operational expenses and maintain consistency across their various AI deployments.

• Enterprises are shifting from AI experimentation to the deployment of "agentic AI," where thousands of autonomous agents operate across multiple systems alongside human staff. • This transition is creating significant operational challenges, specifically regarding rising costs and critical governance gaps in enterprise environments. • While previous industry focus remained on Large Language Models (LLMs) and generative AI, the priority is now moving toward the infrastructure needed to manage these agents.