LFP219 – Automating Intelligence Gathering from GCHQ to Scanning 18bn News Articles for Credit/Reg/Biz-Relevance w/Jeremy Annis CEO Ripjar

Increasing automation is what the tech revolution is about – by definition. In this episode we dive into this topic starting at the highest levels – Government intelligence community quality tech at GCHQ (where Jeremy and his four co-founders had been long-term employees) and the immense value of technology transfer between state and private sectors. We then move through some of the more challenging questions of definition of terms (super-relevant for computers and their usage by business) and that these terms (eg crime, terrorist et al) vary per location – a real challenge for global enterprises.

Finally we move on to a particular speciality of Ripjar (who recently completed a $37m raise) – adverse media screening  – which is not – as it sounded to me – a PR function but rather the automated processing of billions of news articles for credit/regulatory/business relevance re, as it were, early-warning/detection of say credit or regulatory or ethical problems with clients while these are still over the horizon rather than at the front door (or inside it) of the business.

Topics discussed include:

  • bearing crosses – jetlag and karaoke/Christmas parties
  • GCHQ roots and centre of the highest levels of technology and smart folks
  • BBC Micro (computer) and Defender
  • GCHQ one of the three intell agencies in the UK
  • the impact of 9/11 and the “war on terror” – esp at a technology level the nature of the hugely challenging questions of automation of intelligence gathering from internet content
  • defining terms – the importance, slipperiness and relative nature around the world – theory and practical impact for clients
  • adding to the above styles of media writing vary per geography as well multiplying the technological challenge
  • major focal points of crime for Ripjar from the hardcore to the softer end like corporate malfeasance which can be much more relative
  • the realpolitik is that companies need to follow local laws and regulations as well as add on their ethical layer (if they have one lol, not all do…)
  • from a practical perspective re solutions Ripjar give the flexibility for clients to define/search as they wish
  • what is like working at an intelligence agency
  • the multiple challenges of transitioning from a government intell agency to a startup/commercial environment
  • the challenges of a life spent focusing on criminals – from Nietzsche’s “stare long enough into the void and the void stares back into you” to not implicitly as a person seeing the whole of humanity as potential criminals
  • the value of creating a startup in this context
  • intell agencies culture of spinning-out and being open or not – GCHQ and Israel agencies as Case Studies
  • helping governments understand the spread of ISIS propaganda and giving the ability to take down horrible material
  • tracking the emergence of online extreme behaviour
  • the huge range of industries that Ripjar has worked with and how this broadens their perspective
  • the core relevance of the FS industry in terms of setting good practice/leading edge approaches to data and criminality
  • data siloes as a real issue
  • ransomware – including situations where it has been deployed to hospitals or emergency services :-O
  • the connection between some ransomware activities and State-level players
  • detecting and analysing ransomware
  • how data sharing would make this sphere so much the better
  • adverse media screening
  • wide use cases
  • automated reading of news articles and screening them for risk factors for the business
  • early-warning signals in open-source material
  • the 40 year database of 18 billion news articles
  • the challenge of the narrowing of the overton window on leading topics to – in western media at least – key The Current Topics having no diversity of angle
  • broadening this to a bifurcation – Brexit, Trump etc
  • what percentage of practical issues are grabbed by this phenomenon
  • the ability to potentially measure this change in discourse over recent decades
  • probabilistic measure of what might be “the truth”
  • practically-speaking for lower-level credit-related events say there may not have been a narrowing of the window and indeed eg with reporting of court cases there was not much breadth in the first place as that tends to be factual
  • 0.9% of total news articles in the database have relevance to topics Ripjar’s customers care about
  • measuring/modelling a “3D Overton Window”
  • the time-factor in interpreting data and the evolution of data at a practical level for banks
  • fake news
  • Jeremy “fully expects ChatGPT to be used to write media articles in the future”
  • the need to include this in the algorithms/approach and to filter out machines created information
  • commercial shoutouts for RIpjar and Labyrinth their decision-support platform
  • 71% growth in their business in 2022
  • longer term plans

And much much more 🙂

Share and enjoy!