SO I just found another way to use Bing Chat / ChatGPT to do my work for me.

I have had to update our RFFR (Right fit for Risk - an Australian Government addition to the ISO27001 that takes the number of controls from about 117 to about 1,800).

I wrote over 3,800 pages of policies, documents and evidence responses last year for certification and our auditors expect all pages that reference these controls, in all of those 3800 pages, to be labelled in the SOA (statement of applicability).

Every 3 months, the govt adds about 50-150 new controls, so by the time you've updated all the policies and documents, it's a nightmare job updating the SoA because even adding 1 new page in the middle of any document, throws out all the other page numbers.

SO I asked Bing Chat to write me a VBA macro to loop through an array (which contains all the controls, i.e. "1234", "5678", etc.) and then open every document in the folder and find the page numbers that string appears on, then write to a csv with the control number, the page number and the document name.

And it worked, first try. Literal cut and paste (add the controls to the array) and hit F5. And now I have a CSV that lists all controls, and every document they appear in, and the pages in that document.

So it's now a simple case of VLookup, back in Excel, and my SoA is updated and accurate.

That would have taken DAYS to do by hand - and it took maybe 2 hours, start to finish, for me to think of the idea, ask Bing to do it for me, then go and resave the documents to .docm instead of .docx, so Macros can run in them, run the Macro and then do some VLookups to fix the SoA.

In other words, "I will continue to make all the decisions about what I care about (i.e. banning anyone who makes fun of me and spreading insane, rightwing conspiracies) and continue to not pay attention to the rapidly sinking Tesla".

Dude should be the first case study for anyone studying addictions.

So I gave a talk last week at SecTalks on predicting Pump & Dump #scams in #Crypto (you know, other than simply saying "they all are" which would be pretty damn accurate). You know, the ones that run on Telegram groups, where the other group members are the actual suckers.

Process is simple. I took a real world example and analysed the history. Then built an API reading script to crawl for the same signs, across all 2200 tokens in Binance. Basically, you just look for a gradual increase in buy pressure that is close - but not close enough - to baseline. We're not looking for spikes or major changes. We want to see people trying to be subtle.

We do this by comparing the month ave to the day ave and watching for those with a ratio near to 1 but not too near.

Then we look at the purchases when this happens and measure the randomness of them - and look for non-random purchasing. You can see in the Excel based graph the data all has a clear upper and lower bound, this buying was anything but random and was clearly done by a bot.

Then the purchasing tails off and goes quiet.

And that's how we know there's a pump about to happen.