Full Summary
This Thursday morning, multiple sources confirm a surge in investment and innovation focused on securing AI, particularly autonomous agents and code, as companies and governments grapple with new vulnerabilities. Both Ynetnews and PR Newswire report that AI security startups Tenet and NeuralTrust have collectively raised $26 million in seed funding. Tenet secured $6 million to monitor and control enterprise AI agents, while NeuralTrust received $20 million, the largest cybersecurity seed financing for an EU company to date, to identify, secure, and scale AI agents. This funding aims to address the challenge of AI agents accessing corporate systems with limited human oversight. Meanwhile, Help Net Security and SC Media detail AWS Continuum, a new AI-powered platform now in gated preview, designed to manage code vulnerabilities from discovery to fix with minimal human intervention. It uses multiple AI models to reason over a customer's environment, prioritize findings, and suggest remediation. Here's the thing: as AI agents increasingly write code with light human oversight, new risks emerge. Help Net Security highlights a University of Oxford and SaferAI study, revealing gaps in clear ownership for critical control actions like pausing models, and monitoring lags behind AI actions, with OpenAI's review arriving 30 minutes after a session. What nobody expected is the emergence of new AI-specific attack vectors. CPO Magazine reports a "Parameter-to-Prompt Injection" vulnerability, CVE-2026-42824, that turns Microsoft Copilot into a tool for data theft, enabling attackers to steal emails and files. Microsoft has already issued a patch. Additionally, Ynetnews mentions Tenet's research on "Agentjacking," where AI agents are manipulated by attackers. Industrial Cyber confirms Australia's CISC has updated its critical infrastructure rules, mandating risk assessment for AI and legacy systems, and requiring phishing-resistant multi-factor authentication. NAI500 notes cybersecurity leader CrowdStrike's first-ever four-for-one stock split, with shares trading near $694 before adjusting to $173, as the company bets on autonomous AI agents for future growth in a market projected to hit $325 billion by 2030. This means businesses need to immediately update their AI systems, especially those using Copilot, and critically evaluate the security of any autonomous AI agents deployed, as these new vulnerabilities directly impact data privacy and operational integrity.