skip to content

VirusTotal

Multi-engine scanning of files and URLs plus the threat-intelligence and community context attached to a sample. Interviewers ask because reading a report critically — vendor disagreement, generic detections, and the risk of uploading a sensitive file — matters far more than the detection count.

on this pageshow

questions

6

During triage of a suspicious email attachment, when do you search VirusTotal for the file's SHA-256 instead of uploading the file, and what does each action disclose?

level: juniorimportance: must knowfreq 72%

answer

  1. one sends a digest, one sends content
  2. an existing report, or none at all
  3. who can read an uploaded file
  4. no report is not a clean result

basics

~20 s

Search the SHA-256 first: it returns any existing report and sends VirusTotal no file content. Upload only what may be shared, because results reach partners and the community, and file contents can reach premium customers.

solid answer

~40 s

I hash the file on the analysis host and search that first, because a lookup needs only the digest and returns any existing report, with its engine results and first and last submission dates. An upload is a different act. VirusTotal says basic results go to the submitter and the examining partners, reports go to the public community, and **the contents of submitted files may be shared with premium customers**. So I never upload an invoice, a contract or anything naming a customer unless our policy allows it, however suspicious the sender looks. If the hash has no report, I record the file as unknown, not clean: it only means VirusTotal has no public report for that exact file. Then I take it to an internal sandbox or escalate.

code

bash · 2 lines
bash
HASH=$(sha256sum invoice.docm | cut -d' ' -f1)
vt file "$HASH" --include=last_analysis_stats,first_submission_date,names

go deeper

for a junior

Be ready to say why you search the hash before uploading, and to state in one sentence who can receive an uploaded file's contents.

for a middle

Expect to explain what each result establishes, including why an empty hash search is unknown rather than clean, and when an upload is acceptable under policy.

for a senior

An interviewer wants the judgment call on a file that might hold business data, and the route you take instead: an internal sandbox, Private Scanning, or escalation.

for a principal

Be prepared to own the rule for the whole SOC: what may be uploaded, by whom, how an accidental upload is handled, and how you would know the rule is being followed.

## Two actions that look alike VirusTotal's search box and upload button sit side by side, and both end on a report page, so they are easy to treat as the same step. They are not. A **hash lookup** asks whether VirusTotal already holds a report for a file with this digest. An **upload** hands VirusTotal the file itself and starts an analysis. A cryptographic hash such as **SHA-256** is a fixed-length fingerprint of a file's bytes. It cannot be turned back into the document, so searching it reveals that someone is interested in that digest, but not what the file says. VirusTotal's report links accept an MD5, SHA-1 or SHA-256 as the identifier. ## What a lookup gives you If the file is known, the report arrives immediately, without waiting for a scan: - the count of partners that consider it harmful, and each engine's finding and label; - the **first and last submission dates**, which show how long the file has been circulating; - the file names it has been submitted under; - the community score and comments. This is usually enough to move an alert forward, and it costs nothing in confidentiality. Note the report's own analysis date, because a report can be old. ## What an upload sets in motion VirusTotal's documentation describes the sharing plainly: | Who | What they can get | |---|---| | The submitter | Basic results | | Examining partners | Basic results, used to improve their own systems | | The public community | The scanning report, plus comments and votes | | Premium customers | The **contents** of submitted files or pages, through criteria-based searches | The last row is the one that matters on shift. A premium customer can search the corpus and download samples, which is valuable for security research and is exactly why a confidential file must not go in. If it happens by accident, the documented route is a contact form for private uploads, and that is a request rather than an undo button. Anyone may already have seen the file by then. An upload can also reveal your interest to an attacker who watches for their own sample. That is an investigation-tradecraft concern, and it is a second reason to prefer the hash, separate from the data-protection one here. ## A working order in the queue 1. Compute the hash on the analysis host, not on the user's machine. 2. Search the hash and read any report that comes back. 3. If there is no report, classify the file before anything else: could it contain business data, or a customer's data? 4. If it is plainly attacker-built and policy allows it, an upload is reasonable. 5. If it might be confidential, use an internal sandbox, or **Private Scanning** if your organisation licenses it. Private Scanning is VirusTotal's paid route for files that cannot be shared. Uploads are not shared with third parties and are deleted after a retention period, usually 24 hours. The trade-off is explicit: private analyses contain **no antivirus verdicts**, only the output of the other characterisation tools, including sandboxes. Uploading the same file through the standard route as well removes that protection. ## Uploads that do not look like uploads The website is not the only way a file reaches VirusTotal. The documentation lists desktop uploaders, browser extensions and a programmatic API as well, and each one is a submission. That creates some easy mistakes under pressure: - a browser extension or desktop tool that an analyst installed for convenience, which sends a file with one click; - a script or playbook someone else wrote, which uploads when a hash has no report; - a colleague's tip to "just drop it in", given before anyone has asked what the file contains. Find out what your own tools do before you rely on them, because the sharing rules above follow the file and not the button. ## Reading "no report" A hash with no report means VirusTotal has no public report for that exact file. It is not a scan result, because nothing was scanned. Changing a single byte produces a different hash, so a freshly built or per-target attachment will often have none. Record the result as **unknown** and keep working the alert from your own telemetry. A weak answer treats the empty result as reassurance, and a stronger one treats it as a reason to look harder.

  • The user insists the attachment is a harmless customer contract. Does that change what you do?
    It makes me more careful, not less. If the file really is a customer contract, uploading it could expose that customer's data to premium customers, and if it is malicious the hash lookup still works. I search the hash, examine it internally, and upload only if policy allows it.
  • Someone on your team uploaded a confidential document this morning. What happens next?
    I treat it as a possible data exposure and report it through our incident process straight away. We use VirusTotal's private-upload contact form to ask for removal, but we cannot assume nobody saw the file before removal, so the data owner has to be told.

  • A hash with no report on VirusTotal means the file has been scanned and is clean.
  • Uploading a file to VirusTotal only sends it to the antivirus engines.
  • You can delete an uploaded file yourself, so a mistaken upload does no harm.
  • Searching a hash discloses the contents of the document it came from.
open as a page

A file comes back from VirusTotal with zero engine detections. Why is that not enough to close the alert as benign?

level: juniorimportance: must knowfreq 68%

basics

~20 s

Zero detections means no engine flagged the file, not that it is safe. The report may be old, engines lack desktop behavioural detection, and new or targeted samples are often unseen, so the verdict must come from your telemetry.

open as a page

A VirusTotal report shows 4 of 70 engines flagging a file, mostly with generic or heuristic labels. How do you decide what that means for the alert?

level: middleimportance: must knowfreq 64%

basics

~20 s

Treat the count as a lead, not a verdict. Read which engines fired and whether their labels agree, then check dates and file names, and let your own telemetry decide, since VirusTotal only aggregates other vendors' opinions.

open as a page

Before you scan a link from a reported email in VirusTotal, what do you check, and how do you read the URL and domain results afterwards?

level: middleimportance: should knowfreq 52%

basics

~20 s

Check the link for tokens or personal data, because submitted indicators join a dataset others can read. Then read each partner's finding, treat unrated as no opinion, and remember URL and file verdicts are independent.

open as a page

A VirusTotal report on a confirmed malicious file lists contacted domains, IP addresses and sandbox behaviour. How do you use it to scope your estate without chasing noise?

level: seniorimportance: should knowfreq 40%

basics

~20 s

Treat the report as hypotheses from sandbox runs, not your estate. Search first for what is specific to this sample, discount shared infrastructure and routine system traffic, and do not read a quiet sandbox run as proof of inactivity.

open as a page

Your SOC wants VirusTotal lookups in its SOAR playbooks, and analysts already use the free website daily. What policy and licensing do you set?

level: principalimportance: should knowfreq 28%

basics

~20 s

Do not build on the Public API, whose terms bar commercial use. License premium access or drop the integration, then set rules: hash-first lookups, no automatic upload, enrichment rather than detection, and defined failure and accident handling.

open as a page