When performing OSINT research I often utilise Google keywords and advanced features, also known as Google Dorking. It’s a great way to find content and data that a basic search may fail to retrieve. Recently however, I discovered that AI, namely GPT, can retrieve the same results, and sometimes even more, faster than I can dork, prompting me to speculate whether AI has replaced the need to learn advanced OSINT techniques at all. Maybe OSINT 2.0 is learning to master AI prompts to conduct OSINT research faster and better than manually?
I put this to the test. I was recently asked to research a company. I like to use mind-mapping to join up the data that I find. Using Google Dorking and various sites that offer data on public companies I compiled a lot of information in around an hour, all linked together with reference points and a timeline. I then asked GPT to research the same company specifying the parameters. In five seconds it not only retrieved all the information that I had found, but it had additional information that I had not.
I’ll admit that at first I was annoyed that it was faster than me but after thinking about it I realised that it was also very useful. It provided links and evidence along with its findings, which I could follow to verify. The outcome was new sources that I could add to my own playbook. I also decided that I could use AI to assist with the process but that you have to follow a trust but verify rule. Check everything it tells you and back it up with evidence as it may have hallucinated some or all of its findings. It helps to think of AI as a virtual OSINT assistant in this case. You can give it something to research, but make sure you yourself check everything it returns.
As for whether AI has killed Google Dorking, I still use dorking to verify what AI returns as it may of missed something, possibly considering it irrelevant or out of scope.