Radware launches Agent Trust Management for AI agent security

Radware's new solution lets enterprises verify, monitor and govern AI agent activity without resorting to blanket allow-or-block policies.

By The Datatech Times Editorial Team AI level 2 of 5
AI level 2 of 5: drafted by our AI editorial assistant from the company's announcement; fact-checked, edited and signed off by Mark Walker, Editorial Director. What the levels mean
Modern data center control room featuring white consoles with illuminated screens and controls, separated by glass walls from rows of glowing blue server racks under bright recessed ceiling lights.

Radware (NASDAQ: RDWR) has launched Agent Trust Management, a product designed to help enterprises manage the security risks posed by autonomous AI agents interacting with their web applications and business workflows. The product is available immediately as part of the company's Cloud Application Protection services.

The new offering targets a gap that Radware's own threat research has surfaced: according to the company's H1 2026 Global Threat Analysis Report, 77% of organisations are actively deploying or implementing AI agents, yet only 17.2% say they have full visibility into the agents operating within their environments. That mismatch creates a material exposure, particularly in sectors such as financial services, e-commerce and travel, where agents can initiate transactions, access account data and interact with inventory and pricing systems.

How it works

Rather than relying solely on agent identity or observed behaviour, Radware says its system captures an agent's declared intent through active conversation at session level and then continuously evaluates whether subsequent actions remain consistent with that stated intent. Multi-signal identification combines request headers, client-side behavioural signals and direct agent responses to classify both Web Bot Auth-compliant and non-compliant agents; standards-compliant agents are cryptographically verified.

Permission-based governance then allows security teams to set granular controls: an agent can be permitted to browse a catalogue, for example, while being blocked from login or checkout flows. Trust scores evolve as new signals emerge during a session, so an agent that begins behaving anomalously mid-session can be flagged or restricted without terminating the interaction outright.

David Aviv, chief technology officer at Radware, said: "Agent Trust Management enables organisations to establish trust and control what agents can do, so they can participate in the agent economy while protecting critical applications, business processes and data."

Market context

AI agent security is a fast-emerging subcategory within application security, sitting at the intersection of bot management, API security and identity governance. Established players in bot management include Cloudflare, Imperva (now part of Thales) and Akamai, all of which have begun extending their detection and mitigation capabilities to cover AI-driven traffic. Purpose-built agent-security startups have also attracted venture funding over the past 12 months as the volume of non-human API traffic has grown sharply with the adoption of large language model-based workflows.

Radware's framing positions Agent Trust Management as an evolution of its existing bot management stack, which is a credible incremental step rather than a platform pivot. The company's existing cloud delivery and DDoS mitigation customer base gives it a distribution path into enterprise security teams that newer entrants lack. The key competitive question is depth: whether intent-based trust scoring proves more accurate than purely behavioural approaches in real-world adversarial conditions.

Regulatory read-across

The AI agent economy is developing faster than the regulatory frameworks designed to govern it. The EU AI Act's obligations for general-purpose AI systems are phasing in through 2026 and 2027, and there is active debate among standards bodies about how autonomous agent interactions should be logged and audited for compliance purposes. Organisations in regulated sectors, particularly financial services under DORA and healthcare under sector-specific data-protection requirements, will face pressure to demonstrate that third-party AI agents cannot exfiltrate sensitive data or initiate unauthorised transactions. Products that offer session-level audit trails and granular permission controls are well positioned to serve that compliance need, provided they can produce the structured logs that regulators are likely to require.

Radware did not disclose pricing, customer names, or performance benchmarks for the new product. The company's next earnings call will be a natural checkpoint for gauging early commercial traction.

Sector

Topic

Geography