CalypsoAI has highlighted a new era of cyber threats driven by autonomous AI agents, unveiled at Black Hat USA 2025 in Las Vegas on August 5, 2025. These agents, capable of full-cycle threat research, intelligence gathering, and attack generation, have caused a 12.5% drop in security scores across major AI models, signaling a critical shift in the cybersecurity landscape.
CalypsoAI showcases agentic AI as primary cyber threat at Black Hat 2025.
Autonomous AI agents drive threat research and attack generation.
August CASI Leaderboard shows 12.5% security score drop for AI models.
Inference Red-Team uses Signature Attack Packs for vulnerability detection.
New features: Agentic Fingerprints and Outcome Analysis enhance visibility.
Red-Team On-Premises introduced for air-gapped, regulated environments.
CalypsoAI’s findings at Black Hat USA 2025 reveal that autonomous AI agents are now leading cyber threats, surpassing human hackers. These agents autonomously conduct threat research, gather intelligence, and generate attacks at unprecedented speed and scale. The August CalypsoAI Security Index (CASI) Leaderboard reported a 12.5% security score drop across all major AI models, driven by the company’s Inference Red-Team product using Signature Attack Packs to expose vulnerabilities.
The upgraded Inference Red-Team solution introduces Agentic Fingerprints, offering detailed observability into attack agents’ decision-making and exploits. Meanwhile, Inference Defend’s new Outcome Analysis feature provides clarity on why prompts are flagged or blocked, improving response times. As Donnchadh Casey, CEO of CalypsoAI, stated, "As enterprises rapidly adopt AI applications and agents in key functions, it is increasingly important to know exactly what these systems are doing at every step." These advancements empower organizations to address AI vulnerabilities effectively.
CalypsoAI has introduced early access to Red-Team On-Premises, catering to highly regulated industries and air-gapped environments. This solution allows enterprises to deploy agentic red-teaming capabilities in-house, ensuring robust security without compromising speed or coverage. This move addresses the growing need for secure AI adoption in sensitive sectors.
Showcased at Black Hat USA 2025, attended by over 22,000 cybersecurity professionals, CalypsoAI’s solutions highlight its leadership in AI security. Trusted by enterprises like Palantir and SGK, and backed by over $40 million in funding from investors like Paladin Capital Group, CalypsoAI’s innovations, including its Top-Two finalist status at the 2025 RSAC Innovation Sandbox, underscore its role in shaping secure AI deployment.
CalypsoAI’s unveiling of agentic AI as a dominant cyber threat marks a pivotal moment for cybersecurity. With advanced tools like Inference Red-Team and Defend, coupled with on-premises options, the company equips organizations to stay ahead of evolving AI-driven risks, ensuring secure and resilient AI adoption.
CalypsoAI provides the only full-lifecycle platform to secure AI agents and applications at the inference layer, deploying Agentic Warfare™ to protect organizations from evolving adversaries. Trusted by global enterprises including Palantir and SGK, CalypsoAI's industry-leading team of experts is doing the hard miles to ensure security keeps pace with AI innovation.
TechIntelPro empowers B2B technology decision-makers with the latest industry trends, deep insights, leadership perspectives and actionable intelligence to drive both organizational and professional milestones—and thrive on the tech pulse.