AI-Driven Security Risks: Credential Exposure, Shadow AI, and Advanced Phishing
Emerging Technology Risk · 4 sources · Updated 5mo ago · First seen Nov 12, 2025

Artificial intelligence is rapidly transforming the cyber threat landscape, introducing new risks and amplifying existing ones. Recent research revealed that 65% of top AI firms have leaked verified secrets such as API keys and credentials on GitHub, exposing sensitive data and private models to potential compromise. The prevalence of shadow AI—unauthorized or unmonitored AI tools used within organizations—has become a significant blind spot for security teams, as employees increasingly adopt generative AI tools like ChatGPT and Gemini without official oversight. This unregulated use heightens the risk of data leakage and undermines established security controls.
Simultaneously, threat actors are leveraging AI to enhance spear phishing campaigns, targeting IT leaders with highly convincing, context-aware messages that are difficult to detect. Security teams are also adopting generative AI to improve open-source intelligence (OSINT) gathering, using these tools to expand keyword searches and identify evolving threat actor slang. The convergence of AI-driven attack techniques and the widespread exposure of credentials underscores the urgent need for organizations to implement robust secret scanning, monitor shadow AI usage, and adapt defenses to counter increasingly sophisticated, AI-enabled threats.


