New AI guardrails and expanded platform capabilities provide real-time protection across the entire AI lifecycle
SPLX, formerly known as SplxAI, launched its next phase of growth and industry leadership with the introduction of AI Runtime Protection, a powerful new runtime security tool purpose-built for AI applications and workflows. With AI guardrails embedded directly into its platform, SPLX now delivers the most comprehensive end-to-end security platform that protects AI systems across their entire lifecycle, from development to deployment and real-time operation. To showcase the full capabilities of the enhanced SPLX Platform, the company will be hosting 1:1 meetings at Black Hat USA 2025.
“We’re delighted to have SPLX supporting our ongoing journey toward safe and secure AI transformation across our global operations,” said Michael Mosaad, Partner for Cybersecurity Emerging Technologies, Deloitte.
In Q2 2025, SPLX achieved 160% quarter-over-quarter growth, onboarded five new Fortune 500 customers, and delivered a 100% success rate in bake-offs and champion-led POCs for automated, continuous AI red teaming and security testing of AI agents.
Also Read: CIO Influence Interview with Dipto Chakravarty, Chief Product and Technology Officer at Black Duck
“Our rebrand marks a bold new chapter for SPLX – one that reflects our rapid growth, relentless innovation, and clear market leadership in AI security. Over the past year, we’ve seen the AI security landscape shift dramatically, and we’ve evolved alongside it,” said Kristian Kamber, CEO & Co-founder, SPLX. “What started as a pioneering approach to continuous AI red teaming has evolved into the industry’s most comprehensive end-to-end security platform for AI. With the launch of SPLX 2.0, we’re setting a new standard for how organizations build, deploy, and protect AI systems at scale. This transformation positions us to lead the next wave of AI adoption, ensuring that safety, security, and trust are built into every stage of the AI lifecycle.”
“We’re delighted to have SPLX supporting our ongoing journey toward safe and secure AI transformation across our global operations,” said Michael Mosaad, Partner for Cybersecurity Emerging Technologies, Deloitte. “With their pioneering AI red teaming and validation platform, we can accelerate innovation and unleash the full power of AI across our business.”
Deploying AI assistants and agents into production always carries risk, even when extensive red teaming is part of the AI application lifecycle. Malicious users will attempt prompt injections, jailbreaks, and social engineering tactics to make the model leak sensitive data, generate harmful content, or behave outside of its intended purpose. AI Runtime Protection is designed to stop these threats as they occur by continuously monitoring and filtering inputs and outputs. Acting as a real-time firewall for AI, it enables CISOs and security teams to secure AI deployments in production. It enforces strict behavioral boundaries on AI to ensure compliance with internal and external policies, all with near-zero latency, so performance and user experience remain uncompromised.
With the addition of AI Runtime Protection, the SPLX Platform delivers a robust suite of features to secure AI systems in production:
- Real-Time Threat Detection and Blocking: Continuously monitors all inputs and outputs to identify and stop threats as they happen, including jailbreak attempts, prompt injections, and context leaks. SPLX’s dynamic detection engine flags malicious activity in real-time, ensuring AI systems respond safely and stay within intended boundaries.
- Custom Off-Topic Policy Enforcement: With the integrated Off-Topic Guard, organizations can define and enforce policies tailored to their industry, regulatory requirements, or specific use cases. Whether blocking profanity, competitor endorsements, or off-topic conversations, organizations gain precise control over what their AI can and cannot say.
- Deep Attack Visibility: Provides detailed telemetry for every blocked prompt or flagged output. Security teams can drill down into raw input data, identify exact policy violations, and trace the flow of an attempted attack. These insights help organizations refine their defenses and uncover emerging attack patterns across their AI apps and workflows.
- Fine-Tune Guards for Precise Filtering: Precisely fine-tuned guards ensure AI deployments remain fully functional without sacrificing control or security. Security teams can easily adjust guardrail strictness, whitelist known-safe behaviors, and leverage built-in feedback loops to reduce false positives.
SPLX is also introducing Analyze with AI, a new feature that helps security teams cut through red teaming noise and focus on what matters most. The depth and breadth of SPLX’s Probe testing can generate thousands of successful attack paths, strategies, variations, and edge case exploits. For security teams, this can lead to a new kind of complexity: how to prioritize remediation. Analyze with AI is a purpose-built suite of analysis tools that turns raw red teaming output into clear, actionable insight security teams can act on by summarizing, visualizing, and contextualizing results for faster decision-making and dynamic remediation.
As part of its presence at Black Hat USA 2025, SPLX will host an exclusive, invite-only CISO dinner on August 4, 2025, at 6:00 PM PDT at The Cosmopolitan. The dinner will bring together security leaders and executives for a candid discussion on how to manage risk, ensure compliance, and stay ahead of emerging threats in the rapidly evolving era of agentic AI.
Also Read: Scott Holden Joins Vanta as Chief Marketing Officer
