Google Predicts Fully Autonomous AI Cyberattacks Could Emerge Within Six Months

Summary
Google says threat groups are using AI frameworks to automate vulnerability discovery and exploitation, and predicts unsophisticated attacks with little human input within three months and sophisticated autonomous attacks within six.
Key points
- Google Threat Intelligence Group says it has observed threat actors using LLM frameworks to automate vulnerability discovery and exploitation without a human in the loop.
- The article names frameworks including ARTEX, CAIRN, CyberStrikeAI and HexStrike AI.
- Google describes an exploit_agent system used in attacks targeting French media, entertainment and education websites; an exposed attacker staging server contained victim data, reconnaissance information, webshells and brute-forcing scripts.
- Google estimates that attacks requiring little to no human interaction could emerge within three months, with sophisticated fully autonomous attacks likely within six months.
- Google warns automated attacks can shorten defenders’ patching window and generate false leads; it cites a surge of purported, mostly AI-generated PHP2Shell proofs of concept after PHP announced two flaws.
- Google also reports modified Claude Code being used to deliver the SOMBERMEME backdoor in attacks attributed to activity cluster UNC6836, which it suspects is linked to Midnight Neptune.
Article Details
- Event Type
- Reported adoption of AI-powered frameworks to automate vulnerability discovery and exploitation in cyberattacks
- Impact
- Google Threat Intelligence Group reported that attackers are using multi-agent systems and offensive frameworks to find and exploit flaws, including in attacks against organizations in several sectors. Reported consequences include stolen victim data, delivery of the SOMBERMEME backdoor, less time for defenders to patch, and additional workload verifying purported AI-generated exploits. Google estimates that low-interaction attacks may emerge within three months and sophisticated, fully autonomous attacks within six months; these are forecasts, not confirmed developments.
MITRE ATT&CK
Threat Actors
MIDNIGHT NEPTUNEGoogle described UNC6836 as having suspected links to Midnight Neptune; the article does not confirm that Midnight Neptune conducted the reported attacks.UNC6836Google reported attacks by this activity cluster against cryptocurrency firms in North America and Europe using a modified Claude Code app to trigger download of the SOMBERMEME backdoor. The cluster is described as having suspected links to Midnight Neptune.
Malware
Products
Tools
ARTEXIn an advisory last week, the FBI warned that multiple threat actors started weaponizing the ARTEX AI pentesting framework since August and launched attacks against organizations in the financial, government,CairnA notable example is the use of CAIRN, an open-source multi-agent penetration-testing framework, in an attack against the U.S.CyberStrikeAIGoogle says attackers now systematically abuse open-source pentesting suites to run fully automated attacks, including ARTEX, CAIRN, CyberStrikeAI, and HexStrike AI.exploit_agentexample GTIG highlights in its AI Threat Tracker report, hackers used a multi-agent system called ‘exploit_agent’ to fully automate vulnerability finding and exploitation in attacks targeting French media,HexStrike AIGoogle says attackers now systematically abuse open-source pentesting suites to run fully automated attacks, including ARTEX, CAIRN, CyberStrikeAI, and HexStrike AI.
Countries
Francecalled ‘exploit_agent’ to fully automate vulnerability finding and exploitation in attacks targeting French media, entertainment, and education websites.North Koreaaerospace sector, and suspected North Korean hackers using Claude Code to deliver malware.United StatesA notable example is the use of CAIRN, an open-source multi-agent penetration-testing framework, in an attack against the U.S.
Industries
Aerospaceaerospace sector, and suspected North Korean hackers using Claude Code to deliver malware.Cryptocurrencyattacks from UNC6836, an activity cluster with suspected links to “Midnight Neptune,” targeting cryptocurrency firms in North America and Europe, using modified Claude Code.Defensesince August and launched attacks against organizations in the financial, government, technology, defense, healthcare, and other sectors.Educationautomate vulnerability finding and exploitation in attacks targeting French media, entertainment, and education websites.Entertainmentto fully automate vulnerability finding and exploitation in attacks targeting French media, entertainment, and education websites.GovernmentARTEX AI pentesting framework since August and launched attacks against organizations in the financial, government, technology, defense, healthcare, and other sectors.HealthcareAugust and launched attacks against organizations in the financial, government, technology, defense, healthcare, and other sectors.Media‘exploit_agent’ to fully automate vulnerability finding and exploitation in attacks targeting French media, entertainment, and education websites.Technologyframework since August and launched attacks against organizations in the financial, government, technology, defense, healthcare, and other sectors.
