All sane Fintechs will have a security system to try and keep out as many attacks and fraudulent would be clients. However the nature of the attacks and frauds is getting ever-more hardcore. As Martin says in the show back in the day to rob a bank you would have to take significant personal risk when getting into and out of a bank with your shotgun and a high risk of capture by the authorities. These days the risk:return is way lower (and one reason for much higher levels of fraud) as one can sit in a warm office on the other side of the world, far from the local police force and in no physical danger). What is to be done about this massive change of dynamic?
Resistant.AI specialise in adding an additional layer(s) of protection(s) to existing systems and with an approach analogous to antivirus software – namely using AI to detect the latest patterns of attacks and updating their software hourly to keep pace.
As Martin explains there is no such thing as perfect security – however as with most of evolution perfection is not required – merely being one step ahead of the competition. If you make yourself a harder target to defraud than your competitors the fraudsters will got for them not you.
How is this done? What are the challenges? Topics discussed include:
- Resistant.AI has 9 co-founders :-O
- the decision making process amongst so many founders
- the founders prior security startup in 2005 was formed at university and sold to Cisco in 2013
- the emergence of Fintech as encouraging fraud due to the much lower risk of “robbing a bank” outlined above
- the Czech love for American folks songs and the origins thereof in the 1930s and why it persisted through to the 1980s
- why the central Europeans are prominent in security (the prior LFP on security being with a Hungarian)
- technology never solves problems, it only ever shifts them – why?
- implicit vs explicit challenges with secs for security software
- “the curse of any software process is it lacks common sense”
- adding more and more “commons sense” via a system such as R.AIs which builds in ever more patterns of recognition
- the tradeoffs in security systems
- ever stronger security versus minimising friction for clients
- in the US Army when you turn the computer on it takes one hour downloading updated software before it is ready for use.. :-O
- the challenge of infinite information on people leading to more security but more authoritarianism
- comparisons with antivirus software
- the academic theory on the number of loopholes in systems
- the practical solution to this challenge
- the ability to learn very quickly is essential
- case studies of practical issues in security
- South East Asia as the epicentre of the most sophisticated attacks
- “Fintech and Crypto lead to the ability to steal in a scalable way”
- lack of “catching criminals” by the police when it comes to international digital fraud
- the realpolitik of cleaning up your client list – how many of your current clientbase are criminals?
- automated customer risk scoring based on their behaviour cf all-too-often manual processes in banks
- focusing banking staff behaviour of the real catching criminals bit rather than dealing with false AML alerts
- “typically only ~0.1% of the proceeds of financial crime are ever tracked down”
- greater detail in resolution of customers into narrow groups as aiding the ability to detect criminal activity
- tracing the actual flow of funds from fraud
- layered-defence – identity forensics
- how to use R.AI via their APIs
- charging models
- “we don’t want to replace we want to augment and improve”
- “our mission as a company is detecting criminals and stopping them”
- why clients come to R.AI – predominantly losses and too many false alerts that require manual intervention, knowledge – have some clients that one feels uncomfortable about
- shoutouts for what R.AI need right now to be even bigger and better 🙂
And much much more 🙂
Share and enjoy!