Google and FBI Warn AI Agents Now Automate Exploits and Hunt Crypto Wallets

10 minute ago 2 sources negative

Key takeaways:

  • AI exploits targeting Ethereum keystores signal rising custodial risk; holders should self-custody and audit access.
  • Crypto exchanges face deepfake identity threats; investors should monitor exchange security and diversify custody solutions.
  • Unrevoked keys pose systemic risk; watch ETH volatility as AI breaches accelerate in 1-3 months.

Security researchers from Google’s Threat Intelligence Group and the FBI are warning that cybercriminals have moved beyond prompt-based chatbot use and are now deploying autonomous AI agents capable of finding software flaws, writing exploits, and moving through networks without direct human control.

Google’s Q4 2026 AI Threat Tracker, published November 16, documents a multi-agent framework called exploit_agent found on a compromised staging server. The framework connects to several large language models, studies code before and after patches, finds vulnerabilities, and writes working exploits in a continuous loop. Google called it the first fully automated discovery and use of software flaws by malicious actors that it has seen. The campaign linked to that server mostly targeted French websites in media, entertainment, and education, and stolen data was found on the same server.

The report says hackers are using LLMs to compare old and new versions of software updates. When a company fixes a bug, attackers can spot the change and build an attack before many organizations apply the fix. Google pointed to flaws in PaperCut NG/MF software in September as an example. It expects simple autonomous attacks within one to three months and more advanced versions within three to six months.

The Google research also highlights a trend called LLMjacking, in which attackers steal AI credentials and cloud resources, then resell access through underground services such as APIMart and JetNight at lower prices. Other tools named include ARTEX, CAIRN, CyberStrikeAI, PentAGI, ZeroHollow, Agent Zero, and a Chinese-language phishing platform called Jieyou.

In a separate November 13 advisory, the FBI linked a series of intrusions to the ARTEX AI framework. Some attacks moved from initial targeting to privileged access in under three hours. Targets included financial services, government, defence, healthcare and other industries, with personal information belonging to about 67,000 people in South Korea reported stolen.

Microsoft’s October 21 investigation attributed a campaign to Storm-0940, a China-state-aligned group that used ARTEX against hundreds of organizations. After the ConnectWise compromise, attackers attempted sign-ins to more than 2,250 Microsoft 365 tenants, though Microsoft stressed that figure counts attempts, not confirmed breaches.

The crypto-specific risk is especially direct. Barracuda described a November 1 intrusion that lasted about 44 hours and moved through roughly 10 cloud accounts. On compromised systems, the intruder searched for Ethereum keystore files, extended private keys and seed phrase files. Barracuda traced the breach to a cloud access key given to an AI vendor and never revoked, acknowledging: 'That is on us.'

Amazon reported deepfake and altered-document attempts against identity verification at two crypto exchanges between October 3 and November 5. One verification vendor rejected every recorded deepfake attempt, but the campaigns show that automated attackers are now targeting both wallet credentials and exchange identity checks.

For crypto businesses, the common thread is not a novel exploit but trusted access left in place: unretired vendor keys, exposed credentials, unpatched interfaces and leaked application credentials. Because AI agents make each step quick and cheap to repeat, the window between intrusion and exposure of customer funds can shrink dramatically.

Disclaimer

The content on this website is provided for information purposes only and does not constitute investment advice, an offer, or professional consultation. Crypto assets are high-risk and volatile — you may lose all funds. Some materials may include summaries and links to third-party sources; we are not responsible for their content or accuracy. Any decisions you make are at your own risk. Coinalertnews recommends independently verifying information and consulting with a professional before making any financial decisions based on this content.