DorkGPT
FreeMost people who do serious web research have hit the same wall: they know the information they want is somewhere on the internet, publicly indexed on Google, but they cannot get to it with a normal search query. Advanced Google search operators — 'site:', 'filetype:', 'intitle:', 'inurl:' — can unlock that hidden layer of the web, but writing them correctly takes time and practice. Getting them wrong wastes both. DorkGPT solves that problem entirely. You describe what you are looking for in plain language, and the tool converts your request into a precise Google dork query ready to paste directly into the search bar.\n\nDorkGPT is a free web tool built and maintained by PredictaLab, a French research organization focused on disinformation, open-source intelligence (OSINT), and digital investigation. PredictaLab has been operating since 2020, and DorkGPT was created as a natural extension of that work — giving investigators and researchers a faster way to construct the complex search strings they use daily. The tool is powered by OpenAI's API, and the interface is about as minimal as a useful tool can be: a text input field, a generate button, and your results. That is the whole product.\n\nGoogle dorking itself is a legitimate and widely used research technique. A 'Google dork' is simply a search query that uses advanced operators to filter results far beyond what a typical keyword search returns. For example, instead of searching for 'leaked PDFs government', a proper dork might look like 'filetype:pdf site:.gov intext:confidential'. That string tells Google to return only PDF files hosted on government domains that contain the word 'confidential'. Without experience writing these, constructing that string from scratch takes real knowledge of the operator syntax. DorkGPT eliminates that barrier — you type your intent, the AI writes the dork.\n\nThe primary users of DorkGPT fall into several distinct groups. Cybersecurity professionals and penetration testers use it to audit their own systems — checking what information about their infrastructure or organization is publicly indexed and potentially exposed. OSINT investigators use it to surface publicly available data for research, journalism, or due diligence. SEO professionals use dorks to analyze site indexation, find duplicate content, check which pages are indexed, and audit competitors. Academic researchers and investigative journalists use it to find specific documents, datasets, or filings that are buried in institutional or government websites. In each of these cases, the actual searching still happens in Google — DorkGPT only generates the query string, it does not crawl or scrape anything directly.\n\nThe tool also promotes a companion product on the homepage called Predicta Graph, which is described as a more complete investigation tool with custom API actions for fetching posts, comments, and media. Predicta Graph appears to be a separate free product under the PredictaLab umbrella designed for more structured data analysis beyond search.\n\nWhat makes DorkGPT genuinely useful in practice is how it handles query complexity. Multi-operator dorks — combining 'site:', 'filetype:', 'intitle:', and date range filters simultaneously — are difficult to write correctly by hand, especially for someone new to the technique. The AI handles that combination logically based on the intent described. For an OSINT analyst who runs dozens of searches per day, having that done in seconds versus minutes adds up quickly. For a journalist who occasionally needs to find a specific government filing or leaked document, it removes the expertise barrier entirely.\n\nBecause DorkGPT operates on a free, no-login model, it is easy to use without any commitment. There is no account, no API key required from the user, and no visible usage limit surfaced during normal use. The tool is lightweight, fast, and focused. It does one specific thing — generates Google dork queries — and it does that well. For what it is, that is the right call. Adding features that go beyond query generation would likely muddle the tool's clear purpose.\n\nOne practical note worth understanding: DorkGPT is an ethical, legal tool designed for working with publicly indexed information. The Google dorks it generates are intended for use against public search results, not for unauthorized access to systems. That distinction matters — Google dorking is legal and widely practiced; using found information to access restricted systems is a separate legal issue entirely. PredictaLab, which builds tools for fighting disinformation and supporting ethical investigation, is clear about this positioning.