Wednesday, 28 October 2026
1 agent hack today▼ 9 vs yesterday (10)

AI agent finds four severe flaws in EPICS control software

An autonomous hacking agent called Cipher uncovered a critical remote-code-execution bug in open source software used to run particle accelerators and fusion experiments.

By The Agentic Times ·

AI agent finds four severe flaws in EPICS control software
· Image: theregister.com

An autonomous AI agent has discovered four previously unknown vulnerabilities in EPICS, the open source Experimental Physics and Industrial Control System toolkit, including a critical flaw that could let remote attackers run code on systems controlling particle accelerators and fusion experiments. The US Cybersecurity and Infrastructure Security Agency issued an advisory on 27 October 2026, and users are told to update to EPICS Base version 7.0.11, which contains the fixes, according to The Register.

The most serious bug, CVE-2026-114595, is an integer overflow rated 9.8 under CVSS v3.1. It leads to an attacker-controlled out-of-bounds heap write and potentially remote code execution. The other three are rated high severity: CVE-2026-114596, an improper message-length validation flaw rated 8.2 that causes an out-of-bounds memory write and crashes the controller process; CVE-2026-114597, an insufficient bounds-checking bug rated 7.5 that allows out-of-bounds memory reads and process crashes; and CVE-2026-114598, a type confusion flaw also rated 7.5 that can cause denial of service and may expose adjacent memory contents. All four affect EPICS Base.

"Successful exploitation of these vulnerabilities could allow an attacker to cause a denial-of-service condition, corrupt memory, disclose sensitive information, or potentially execute arbitrary code," CISA said in its advisory, as reported by The Register.

The agent, called Cipher, found and validated all four issues in three hours and 49 minutes while investigating a system it had not seen before, Causal Security co-founder Shubham Raj told The Register. "The initial investigation required no human intervention, followed by one instruction authorizing deeper exploitability testing in a secure environment," Raj said. He described Cipher as "a fully autonomous super hacker that can reason across the technology stack". Earlier in the year, according to The Register, the same agent found multiple critical attack paths in power-grid systems used across North America and Europe, prompting a CISA advisory on 8 October.

EPICS was originally developed by Los Alamos National Laboratory and Argonne National Laboratory. It is used to build and remotely control distributed SCADA systems, which link operators to physical devices. The Register reports it is in use at around 90 major scientific, university and industrial projects and facilities worldwide, including Argonne's Advanced Photon Source and ITER's fusion reactor control system.

Exploitation requires network access to the vulnerable EPICS services on a facility's control network, Raj said. That could happen if networks were not properly segmented, or if another system had a route into the EPICS network. "Once an attacker is able to find a network, which is either misconfigured or they are able to reach, it's pretty easy to carry out the attacks," he said. He added that attackers would not need deep operational technology knowledge: "Someone with conventional exploitation skills could investigate them, and increasingly capable AI systems could help develop exploits without requiring specialized OT expertise."

Raj set out one scenario for the critical bug. "Imagine a particle accelerator in the middle of an expensive experiment. An attacker could potentially crash a controller, interrupt critical measurements or control functions, and force operators to halt the experiment. That could mean lost experimental work, significant downtime, and expensive recovery." A working remote-code-execution exploit, he said, could let an attacker "take control of the affected process, manipulate readings, interfere with control functions, or disrupt the equipment connected to it".

Manipulating specific physical processes still demands extra knowledge, Raj said. "But an attacker doesn't need to understand the physics of a particle accelerator to disrupt the software controlling it. What's changing is that attackers may no longer need years of specialized expertise just to discover these weaknesses."

Sources

  1. theregister.comhttps://theregister.com/security/2026/10/27/ai-hacker-finds-4-severe-bugs-in-popular-industrial-control-software-in-under-4-hours/5305811