Black Hat USA 2026: Vendor Announcements Roundup, Part 2

Security vendors used Black Hat USA 2026 in Las Vegas to announce new products, platform updates, research and initiatives spanning vulnerability management, artificial intelligence, data protection and managed security. The following highlights continue SecurityWeek’s event roundup.

AI, exposure management and application security

• Astelia introduced agentic AI features for its exposure management platform. The capabilities are designed to analyze vulnerability reachability, assess operational consequences, coordinate remediation and track issues through resolution. Human approval remains required at key stages, with actions recorded for audit purposes.
• Drata announced limited availability of AI Agent Governance for its Trust Management Platform. Initially supporting Anthropic, the offering is intended to help organizations identify, monitor and govern internal AI agents while maintaining traceability.
• Horizon3.ai expanded NodeZero with autonomous web-application penetration testing. The company says the feature identifies exploitable attack paths, estimates their business impact and links them to known adversary tactics. The announcement follows Horizon3.ai’s disclosure of a $250 million funding round.
• Legit Security released VibeGuard 2.0, an endpoint security tool for AI coding assistants including Claude Code, Cursor and GitHub Copilot. Updates include controls for agent commands, Model Context Protocol security, monitoring policies, protections against risky operations and anti-tampering measures.

Data, recovery and endpoint protection

• AvePoint added Kinetic Classification to its Confidence Platform, continuously reassessing data sensitivity across Microsoft 365, Google Workspace and other applications. New Rapid Recovery features include a recovery wizard and Express Recovery for Entra ID to support restoration of priority data after incidents.
• Huntress made RMM Guard available at no additional cost to customers with a deployed agent. The capability detects and blocks unauthorized remote monitoring and management tools, supports allowlists and blocklists, and adds controls for isolated systems. The release comes as a vulnerability in N-central RMM is being exploited.
• Netskope announced the Netskope One DataSec Command Center, a centralized control plane intended to discover, track and protect sensitive information across applications, networks and AI environments.

Threat research

CrowdStrike reported that AI is increasingly incorporated into criminal operations, including vulnerability exploitation, cloud attacks and supply-chain compromises. Separately, Cisco Talos described how threat actors use large language models to develop malware, build fraud infrastructure and accelerate vulnerability research, often without requiring advanced jailbreak techniques.

ProjectDiscovery announced general availability of Neo v1 and a pay-as-you-go option, positioning the platform as a way to support continuous, automated security testing across code, applications and APIs.