Darktrace Advances Behavioral Security to Protect AI Agents, People and Infrastructure
Darktrace confirmed no compromise after LockBit claimed a breach, stating its systems and customer services remain secure. The company also unveiled the Darktrace Behavioral Defense Platform to protect AI agents, people, and infrastructure using Adaptive AI for real-time threat detection and response.
Darktrace stated that following claims by the cyber-criminal gang LockBit of a breach into its internal systems, a full security review found no evidence of compromise. The company confirmed its services remain operational and customer data fully protected, with no links to compromised data in LockBit’s posts. Darktrace emphasized continued monitoring while asserting confidence in its systems' security based on current investigations.
The newly introduced Darktrace Behavioral Defense Platform aims to secure modern enterprises as AI systems increasingly interact with people and critical infrastructure. Built on over a decade of behavioral security innovation, the platform uses Adaptive AI to continuously learn and detect deviations in behavior across AI agents, trusted identities, and connected systems. It provides unified visibility and autonomous response to mitigate risks before they escalate, addressing threats that static rules or signatures may miss.
A key component, Darktrace / SECURE AI, focuses on securing AI deployment, management, and development alongside a new Platform Portal offering a 360-degree view of insights and threats. The portal correlates real-time data across AI, people, and infrastructure, helping security teams prioritize risks. Darktrace noted a 13% rise in average AI service connections among customers in the first half of 2026, underscoring the growing complexity of securing diverse AI interactions.
NIST’s June 2026 proof highlighted the limitations of fixed AI guardrails against adversarial prompts, reinforcing the need for continuous monitoring. Darktrace’s mid-year threat update found attackers increasingly exploiting trusted identities and AI services rather than software vulnerabilities. The platform’s Adaptive AI builds unique behavioral profiles for each organization, enabling real-time detection and response to novel threats, including rogue AI agents or misconfigured systems, without relying solely on static indicators.