technology

Microsoft’s JPMorgan System Encounters Issues; Google: Era of Autonomous Attacks Has Arrived

Published November 17, 2026 · Updated November 18, 2026 · 4 min read

Amid a wave of technical disruptions affecting major companies such as United Airlines, JPMorgan Chase, and Microsoft, Google has released a significant threat intelligence report detailing alarming developments in cybersecurity. The report, published on November 16, raises the warning that AI-driven cyber threats have crossed a crucial threshold, ushering in what it describes as the era of “autonomous attacks.”

The report titled “GTIG AI Threat Tracker: Agentic Threat Activity Crosses a Significant Threshold,” issued by Google Cloud, outlines observations made by the Google Threat Intelligence Group (GTIG) in collaboration with its cybersecurity partner, Mandiant. For the first time, the researchers detected hackers deploying autonomous AI agents capable of operating without ongoing human intervention. This advancement marks a full automation loop that covers everything from identifying software vulnerabilities to crafting exploit code and executing attacks.

Google explicitly named several threat groups suspected of having state affiliations that are increasingly misusing generative AI technology. For instance, organizations such as LEAKY CASTLE, which is believed to have connections to China, have been found to harness large language models for the automation of analysis, translation, and interpretation of stolen victim data.

Another notable organization, codenamed MIDNIGHT NEPTUNE, has been identified for employing deceptive practices such as creating false recruiter profiles and exploiting Microsoft Copilot’s name to generate GitHub personal access tokens (PATs). These tactics provide cover for their lateral movements and infiltrations.

The report also highlights the significantly reduced time gap between software patches and vulnerabilities—a trend researchers observed when they discovered a multi-agent attack framework called exploit_agent on a compromised attacker infrastructure server. This tool can automatically analyze security patches publicly released by software companies. By comparing code changes before and after updates, it can reverse-engineer the location of vulnerabilities and autonomously generate and run exploit programs. This framework has reportedly targeted French media, entertainment, and educational websites to steal sensitive data.

Moreover, between late August and early October 2026, GTIG monitored threat actors employing a custom attack orchestration platform based on the open-source penetration testing framework, PentAGI, to conduct systematic automated reconnaissance and vulnerability scanning against organizations across Asia.

Google underscored the gravity of the situation in its official statement on Monday, declaring, “The era of autonomous attacks has arrived.” The company emphasized that defending against machine-speed threats at human pace is no longer feasible. Google indicated that cybersecurity defenses must similarly advance towards high levels of automation to effectively respond to this new normal.

In the week prior to Google’s dire warnings, the U.S. aviation, banking, and technology sectors experienced a string of technical failures that exposed the vulnerabilities of crucial digital infrastructure.

On November 16, United Airlines faced a technical glitch that temporarily delayed flights, resolving the issue around 6 PM EST. According to Reuters, United recorded a total of 645 flight delays and 25 cancellations on that day—though not all of these were directly attributable to the malfunction.

Just a few days earlier, on November 13, the Federal Aviation Administration (FAA) encountered a communications overload due to a router failure post-network upgrade, which resulted in a nationwide halt of flights impacting thousands. Both U.S. Transportation Secretary Sean Duffy and United Airlines officials stated there was no evidence of malicious cyber activity involved.

In the banking sector, JPMorgan Chase acknowledged that a software flaw had led to delays in some trade processing, but confirmed that customers could still access their funds without issues. The bank reported no signs of malicious attack. Similarly, Bank of America and Wells Fargo customers experienced related service outages on the same day, while U.S. Bancorp issued warnings regarding potential delays in interbank payments.

Microsoft, too, faced a global system failure on November 13 due to server issues, which impacted core services like Outlook, Teams, and Copilot for several hours. The company stated there was no indication of malicious activity associated with the outage.

As of now, neither Google nor the affected companies and regulatory bodies have confirmed any direct connection between the recent spate of technical failures and emerging AI-driven cyber threats. Nevertheless, analysts point out that amidst the dual challenge posed by increasingly complex corporate digital infrastructures and attackers now operating at machine speed, enhancing the technological resilience and automated defense capabilities of essential infrastructures has become an urgent priority.