Definition
Remote code execution is Remote code execution (RCE) is a security flaw that lets an attacker run their own code on someone else's computer or server over a network, often with no login, which makes it among the most severe kinds of bug.
Remote code execution, or RCE, is a vulnerability that lets an attacker make a computer run instructions of the attacker's choosing, from somewhere else, across a network. The computer might be a web server, a phone, an AI agent's host machine or a developer's laptop. Once an attacker can run code there, they can usually read files, steal passwords and keys, install other software or use the machine to reach more systems.
Security teams treat RCE as near the top of the severity scale because the bug hands over control, not just a peek at data.
The word that changes the risk most is "unauthenticated." An unauthenticated RCE can be triggered before any login, by anyone who can send a request to the target. An authenticated one needs a valid account first, which narrows who can use it. Severity scores such as CVSS reflect this: a flaw that needs no privileges and no user interaction scores much higher than one that needs both.
RCE rarely means someone "hacked in" by guessing. It usually comes from a mistake in how software handles input:
The fix is almost always an update from whoever maintains the software, which is why patch speed matters.
Two things connect RCE to AI work. First, AI models are now used to find these flaws, which shortens the time between a bug existing and someone locating it. Second, AI agents that run code or browse the web can be tricked, through prompt injection, into running something harmful. Running agents inside an agent sandbox limits how far a successful attack can reach.
A penetration test is an authorized attempt to find flaws like RCE before attackers do, and red teaming applies the same mindset to AI systems.
A data breach means information was exposed. RCE is a way in: it lets an attacker run code, which can then be used to cause a breach, install malware or take over the machine.
It means the attacker needs no account or password to trigger the flaw. Anyone who can reach the vulnerable system over the network can try it.
Prosecutors charged OpenAI engineer Abdulrahman Alfozan with planting a recording device in an executive conference room, citing prompts about building one from a linked chatbot account. Your chat history is a record a court can read.
PHP released fixes on October 21 for two flaws rated 9.2 that let attackers take over servers without logging in, found by Wiz with Gemini and GPT models. If your site runs PHP, updating today is the job.