Aug 8, 04:17 PM

Black Hat 2025 Highlights CrowdStrike Report of 1.8 Million Attacks, 62,000 Breaches on AI Agents with 1.5% Failure Rate

At the 2025 Black Hat cybersecurity conference in Las Vegas, experts highlighted the increasing threat posed by artificial intelligence (AI) tools being weaponized by cybercriminals. A report from CrowdStrike warned that threat actors are leveraging AI to scale operations and accelerate attacks, particularly targeting autonomous AI agents integral to enterprise automation. Researchers from Carnegie Mellon University revealed alarming findings from a red-teaming exercise, documenting 1.8 million attacks and 62,000 breaches, with even the most hardened AI agents failing 1.5% of the time. Security specialists emphasized that relying solely on policy to secure AI agents is insufficient, as vulnerabilities in bad code, malicious models, and rogue agents persist. The conference also showcased how AI is now both a tool for attackers—used to create convincing fake résumés, LinkedIn profiles, and deepfakes for video interviews—and a defense mechanism. These developments illustrate a cybersecurity landscape that is increasingly complex, with AI playing a central role in both offensive and defensive strategies.

#Black Hat #Las Vegas #CrowdStrike #Carnegie Mellon University #LinkedIn

Written with ChatGPT (GPT-4).

Sources

Additional media

Image #1 for story black-hat-2025-highlights-crowdstrike-report-1-8-million-attacks-62000-breaches-80e8efa9

Image #2 for story black-hat-2025-highlights-crowdstrike-report-1-8-million-attacks-62000-breaches-80e8efa9

Image #3 for story black-hat-2025-highlights-crowdstrike-report-1-8-million-attacks-62000-breaches-80e8efa9

Black Hat 2025 Highlights CrowdStrike Report of 1.8 Million Attacks, 62,000 Breaches on AI Agents with 1.5% Failure Rate

Sources

Additional media

Similar Stories