Google flags AI-built zero-day exploits
Google says it disrupted a criminal group that was using AI to identify and weaponize a previously unknown software vulnerability, marking another signal that AI is speeding up offensive cybersecurity work. The report comes from Google Threat Intelligence Group and focuses on how threat actors are moving beyond simple AI-assisted productivity into active use of AI for reconnaissance, phishing, malware development, and exploit research. The specific target and model were not disclosed, and Google said the model used was likely not Gemini or Anthropic’s Claude.
This is less a product story than a security warning from a major platform vendor, but it matters because it shows AI is now part of the exploit pipeline, not just the spam pipeline.
- –The notable shift is AI-assisted vulnerability discovery, not just AI-written phishing or malware copy.
- –Google’s disclosure is credible because it comes from GTIG and is tied to a disrupted real-world campaign, not a hypothetical.
- –The immediate impact is on defenders: faster attacker iteration means shorter response windows for patching and detection.
- –This is not a consumer-facing launch, so the main “product” angle is Google’s threat intelligence reporting and detection capability.
DISCOVERED
2h ago
2026-05-11
PUBLISHED
5h ago
2026-05-11
RELEVANCE
AUTHOR
forbes