Featurespace

What does it mean for a fraud model to be fair? Featurespace’s Innovation Lab answers that very question in its new paper, published this week in collaboration with Cambridge AI ethicist Dr Eleanor Drage, and PhD student Yulu (Niki) Pi. It’s the first ever systematic study of algorithmic bias for a transaction fraud model, also covering how this fairness can be measured, and how the measurement is impacted by class imbalance. The paper finds that fairness-through-unawareness does not guarantee unbiased fraud detection. As a result, it’s vital that those working across the anti-fraud ecosystem more actively measure and mitigate biases. Give the paper a read here: https://lnkd.in/eZeQeHpN

Check out this fantastic opportunity in **Atlanta** at Featurespace: Senior Project Manager

How should legal and financial crime compliance teams balance regulation and technology? Steve Goddard, Fraud Subject Matter Expert at Featurespace, speaks to Tenet 's specialist fraud solicitor Arun Chauhan - FICA, Prof.PgDip (FCC) at 12pm BST on Tuesday 10th, on readiness for upcoming regulatory deadlines, anticipating future risks, and defining what 'good' looks like. Register for the webinar here: https://lnkd.in/eGQ2pa2Y

Based on news reports in the last 24hrs ⏲ , it looks like the PSR is going to drop the maximum limit on APP fraud reimbursement from £415k to £85k for when it comes into force in October in the UK 📅. The limit has been the focus of many discussions in the last year, but will the reduction have much of an impact and on who? Here is a list of who may be impacted and how: - Existing CRM participant and fraud guarantee institutions 🏛 -> little impact as already refunding over £85k - PSP with strong controls 🏦 -> reduces risk of low volume but high value claims that have a disproportionate impact due to "bad luck" - PSP with immature or weak controls 💷 -> risk is still high of multiple claims of high values although single cases may be of lower value. - Victims 😢 -> More than 99% are not impacted but a very small number of victims that lose life changing sums will still be heavily impacted, although this is better than not being reimbursed at all. - First party fraudsters 🤥 -> Little impact as likely to be focused on lower amounts to avoid suspicion. - Organised scammers 😈 -> no impact, as reimbursement is immaterial to them. Mules 🐴 -> no impact as accounts still likely to be closed, losses written off and markers added to credit files. If there is limited impact, then will lobbying for amendments continue? I think focus should still be on having customer processes ready and improving scam detection based on data driven approaches rather than focusing on limits.

Missed out on Oasis tickets this weekend? Beware: The resale is a fraudsters' playground. Snap up tickets with confidence using our top tips below. They're based on a conversation our President of the Americas Carolyn Homberger recently had with PYMNTS. Take a look here: https://lnkd.in/ggySb863 #Oasis #tickets #scams #Featurespace

Thank you, Datos Insights, for an incredible day 1 of #FinCrimeForum yesterday. The atmosphere was great, the sessions were insightful, and the conversations were inspiring. 👌 We especially enjoyed seeing Shelby Archer from SentiLink and Carolyn Homberger from Featurespace – thank you for stopping by! 👏 📍 If you're here for day 2, the Entersekt team can't wait to meet you. Visit us at booth #6.

Great insights from our Head of Product Dondi Black and panelists from Deluxe and Featurespace into how technology can keep #payments safe at Fintech South. Thank you to everyone that attended their session, and thanks to the Technology Association of Georgia for hosting a wonderful event!

Imagine if there was a single source for scam intelligence across payment rails... That's the recommendation of the Scams Information Sharing industry work group, which includes our founder David Excell. Its new report, released today, outlines that the industry should consider a solution or establish an independent information exchange framework to provide a single source for scam intelligence 🎯 This could help detect scams more quickly and avoid potentially devastating losses. It's vital work as we continue to make the world a safer place to transact. Learn more and read the full report here: https://lnkd.in/gcWutWUu #fraud #fincrime #scams #paymentsindustry #Featurespace

Join us on Tues and Weds at Datos Insights' Financial Crime Forum in Charlotte, NC. At Booth #1, we'll be ready to talk AI, beating APP scams, and how our new collaboration with OrboGraph is set to turn the tide on check fraud - once and for all. Come along and say hi! Not going to make it? Find out more about how we make the world a safer place to transact here: https://lnkd.in/gir4c9u #FinCrimeForum #TechforGood #Scams #Fraud #Featurespace 

We are devastated by the loss of Mike Lynch. Our thoughts and condolences are with his family and loved ones, and all the others affected by this dreadful tragedy. It is a high statistical probability that Featurespace wouldn’t be a thriving technology company without Mike. Our co-founders, Professor Bill Fitzgerald and Dave Excell, were inspired by Mike’s combination of intellect and commercial acumen. They benefited from his friendship and guidance. Mike’s personal investment in Featurespace supported the development of an invention that has proved fundamental in the progress of AI and he served as a Non-Executive Director for over 10 years from 2008 - 2019. Mike was a true champion of the UK technology sector, including the need for greater diversity, and advocated for many female leaders – including our CEO Martina King. Cambridge, the technology sector, colleagues and friends will be unable to calculate the loss of this truly extraordinary, intellectual, inventive, humorous and generous man. He will be sorely and deeply missed.