The rise of autonomous AI agents is reshaping the cyber threat landscape, moving fraud and cybercrime far beyond traditional phishing and social engineering. What once required constant human direction can now be partially executed by AI systems capable of reconnaissance, credential theft, exploit generation, and data exfiltration with minimal oversight.
Recently, several incidents have shown how quickly AI-driven cyber threats evolve. For example, reports linked Anthropic’s Claude Code to cyber-espionage activity, while another breach allegedly used a jailbroken AI system to steal over 150GB of data. As a result, AI is no longer just assisting cybercriminals but increasingly acting as an operational layer within attacks. Meanwhile, AI agents can execute tasks continuously at scale, allowing attackers to weaponise the same systems enterprises use in daily operations.
The challenge is not purely technical. Governance, interoperability, privacy protections, and cross-platform standards all remain unresolved. However, as AI agents become more autonomous, the absence of trusted identity frameworks may become one of the largest structural weaknesses in cybersecurity.
Ultimately, the rise of AI-driven cybercrime signals a larger transformation in digital security. Organizations can no longer rely solely on reactive defences or post-incident investigations. In an era of autonomous agents, trust, provenance, and verifiable identity are becoming foundational requirements for keeping fraud manageable at scale.
Source:
https://www.techradar.com/pro/ai-agents-now-commit-and-conceal-cybercrimes-on-their-own
