- Connect using official plugins or MCP servers using OAuth and Google authentication.
- Differentiation between queries of existing hashes and uploading of unknown files to manage quotas.
- Using AI to decode obfuscated scripts and automate the understanding of malicious code.
- Risks associated with the privacy of shared data and the use of AI by cybercriminals.
Have you ever wondered if it's possible to combine the brains of ChatGPT with the power of VirusTotal to analyze threats without leaving the chat? Well, it turns out it is, and it's no joke. This alliance allows anyone, from a professional cybersecurity analyst to someone who's simply worried they might have downloaded a suspicious file, to investigate domains, IPs, and files in a much smoother and more conversational way.
The most amazing thing is that you no longer need to be jumping from tab to tab endlessly copying and pasting hashes. Thanks to modern integrations like the Model Context Protocol (MCP) , artificial intelligence can query threat databases in real time, helping us decipher obfuscated code or understand why an antivirus has flagged a file as dangerous—all while we chat with the AI.
How to connect VirusTotal to ChatGPT step by step

To get started, the most straightforward option is to use the official VirusTotal plugin . The process is quite simple: first, access the plugin's installation link and click "Install." Then, the system will ask you to authenticate via Google through VTAI (VirusTotal AI). Review the permissions and grant access. Once this is done, simply open a new chat, type the @ symbol , and select VirusTotal to begin making requests, such as requesting a report for a specific domain.
If you prefer full control or work in an environment that supports Developer Mode , you can create a custom application. In this case, go to ChatGPT's settings, enable Developer Mode, and create an application called VirusTotal. The crucial step here is to enter the MCP server URL (https://ai.virustotal.com/mcp) and select OAuth as the authentication method. It's important to note that there are query limits , typically around 60 per minute and 1000 per day, to prevent system abuse.
Another interesting alternative is using Composio . To do this, you install the Composio plugin, authenticate on their platform, and authorize the connection with VirusTotal. This way, you can use commands like @Composio to summarize VirusTotal messages or alerts directly in the OpenAI interface.
File analysis and quota management

When you attach a file to ChatGPT and request an analysis, the connector doesn't upload the file blindly. First, the tool calculates the file's hash to see if VirusTotal already recognizes it. If the hash already exists in its database, it simply retrieves the previous report , which uses one query from your daily quota. However, if the file is completely unknown, then the document is uploaded —a process with its own contribution limits, independent of search queries.
It's crucial to understand that not everything without detections is safe . An empty report or a pending analysis doesn't mean the file is harmless; it simply means the system doesn't have enough data. Furthermore, you must be very careful with data privacy , as anything uploaded to VirusTotal is shared with the security community. Therefore, you should never upload internal company code or personal data without explicit permission.
Boosting malware analysis with AI

Beyond simply connecting, ChatGPT is a powerful tool for accelerating reverse engineering . For example, if you encounter a PowerShell script that looks like hieroglyphics, you can ask the AI to take on the role of a malware analyst and explain what the code does. This is especially useful for identifying hidden malicious URLs or understanding the logic of an HTA file that attempts to execute commands on the system.
In more complex cases, where malware uses steganography to hide data within PNG images or splits PDF files into small chunks to evade detection, ChatGPT can help us write Python routines to decode those strings in Base64. While the AI can't execute the code for us (for obvious security reasons), it gives us the exact roadmap to do so ourselves in a secure, virtualized environment.
The dark side: AI as a weapon for attackers

It's not all good news, as cybercriminals are also using these tools. It has been shown that it's possible to bypass ChatGPT's ethical filters to create malicious code snippets that, when combined, form a zero-day exploit. Some researchers have even managed to develop malware that goes undetected by most VirusTotal engines using obfuscation techniques suggested by the AI itself.
From creating hyper-realistic phishing emails that don't contain a single spelling mistake, to impersonating public figures for cryptocurrency scams, AI is lowering the barrier to entry for attackers. The ability to automate vulnerability scanning and scale spam campaigns makes the threat landscape much more aggressive and faster-moving.
Tips to stay safe
To avoid falling into AI traps, the first step is not to blindly trust chatbot responses, as it may fabricate data (hallucinations). It's vital to keep software updated, use multi-factor authentication (MFA) , and have an active firewall. Furthermore, for those handling very large files that exceed VirusTotal's upload limit, it's best to analyze the individual hashes of the internal files instead of attempting to upload the entire compressed package.
The integration of VirusTotal into ChatGPT transforms threat research, making report retrieval and code analysis much faster and more understandable. While AI can be used to create more sophisticated malware, it also empowers defenders with the ability to react quickly, provided human judgment and caution are exercised when handling sensitive data.