AI · 28 giorni fa
GTIG di Google ha visto il salto che cambia la lettura del problema: nel secondo trimestre 2026, un singolo accesso cloud è bastato per avviare e portare a termine in meno di sei ore una campagna di mass credential harvesting. Separatamente, GTIG attribuisce a UNC6780 l’uso di AI coding assistants e scanner LLM manipolati per compromettere supply chain open source.
Il meccanismo è semplice e più veloce del controllo umano: più agenti AI si dividono il lavoro, correggono gli errori da soli, continuano a scandire e a riprovare, e comprimono in una singola sessione ciò che prima richiedeva giorni. Il risultato non è il modello in sé, ma la velocità con cui una compromissione cloud o un ambiente di sviluppo assistito può trasformarsi in furto di credenziali o in un package avvelenato prima che una verifica normale intercetti la manomissione.
Per chi usa AI assistita su codice, ticket o dati cloud, il punto è che la finestra di reazione si accorcia fino a diventare un problema operativo. Il rischio si sposta così dalla sola governance dell’uso non approvato all’abuso diretto di workflow agentici che possono aggirare i tempi su cui SOC, identity e supply-chain defense fanno ancora affidamento.
20 fonti che coprono questa storia
Why 2026 is the Year to Upgrade to an Agentic AI SOC
Agentic AI SOCs differ from copilot-only models by autonomously prioritizing attacks over alerts, executing closed-loop containment, and providing traceable reasoning for every decision, allowing anal...
Proofpoint SOC Analyst Agent Uses OpenAI Cyber Models
Proofpoint’s SOC Analyst Agent uses OpenAI Daybreak models to help security teams investigate threats, connect signals, and automate recurring analysis.
This AI threat update provides GTIG’s findings on adversarial misuse of AI including Gemini and other non-Google tools.
Hackers build AI frameworks for widescale credential theft
Threat actors are increasingly switching from AI-powered coding assistants to multi-agent frameworks that automate every stage of an attack.
NCSC Warns Shadow AI Creates New Security Risks
NCSC warns unapproved AI tools can expose corporate data and create new security risks
Understanding why staff use unapproved AI tools is key to managing the security challenges they can create.
How to secure edge AI in customer-owned environments | Microsoft Security Blog
As AI moves into customer-owned environments, organizations need new ways to verify the systems, software, and AI assets they trust before releasing sensitive data, credentials, and models.
AI Coding Agents Are Installing Unknown/Untrusted Code on Corporate Networks - Schneier on Security
We cannot forget that AI coding agents are not yet trustworthy: Researchers at a stealth startup in Israel scanned 6,214 live domains belonging to defense contractors, Fortune 500, and Big Tech companies. Of the 8,265 llms.txt and llms-full.txt files they found (many sites hosted both an llms.txt and an llms-full.txt file), 120 of them, each on a different site, pointed to one or more code packages or domain names that weren’t registered. To test what happens when an AI agent processes such files, the researchers registered a handful of the unclaimed names and hosted packages that caused any machine executing them to reach out to their server. Within an hour, the researchers received a phone-home response from a Fortune 500 company. Over time, they got a few dozen more, some from more Fortune 500 companies and others from startups. Their beacon also recorded the chain of parent processes that spawned each install, ultimately revealing that coding agents, including Claude, OpenAI’s Codex, and Nous Research’s Hermes, were involved. Anthropic, OpenAI, and Nous Research did not respond to requests for comment by the time of publication...
Why judgment is emerging as cybersecurity’s defining skill
AI speeds up security analysis, but automated actions without human context risk massive operational outages. Here is how CISOs can safely govern AI autonomy.
Booz Allen reports AI models approaching autonomous cyberattack capability as critical infrastructure response windows narrow.
A five-part inventory for your AI agent credentials - Help Net Security
Roy Katmor explains the review gap around AI agent credentials and what teams should inventory before an agent acts on its own.
AI 'Machine Speed' Cuts 2-Week Attack Down to 10 Hours
The incident demonstrates how frontier AI agents can dramatically compress an attack timeline and coordinate a large-scale breach.
Part of the PlainSec briefing for 2026-09-02