Google Warns of Hackers Using AI to Create Working Zero-Day Exploit

GTIG’s Q2 2026 report warns that both cybercriminal syndicates and nation-linked APTs are industrializing generative AI to create and scale offensive operations: analysts found a likely AI‑generated Python zero‑day exploit that bypasses 2FA, the PROMPTSPY Android backdoor that integrates Gemini for autonomous device control and data theft, supply‑chain compromises of popular developer tooling (exfiltrating API keys and CI/CD secrets), and multiple APTs using LLMs to discover and operationalize vulnerabilities; GTIG recommends auditing CI/CD pipelines, tokens, and AI dependency chains.