IBM has unveiled detail of its much-anticipated Telum Processor, which has built-in AI inferencing acceleration technology and can be used for real-time apps such as fraud detection. The idea is that the chip will be able to flag up a scam while the transaction is actually taking place.
► Will help companies move from fraud detection to fraud prevention
► On-chip AI can be 100% devoted to AI workloads
While the key area of adoption is expected to be banking and finance, the chip may also be used to guard against fraud in almost any sector. As well as fraud prevention, it could be used for clearing and settlement of trades, anti-money laundering and risk analysis. The first Telum-based systems are expected in the first half of next year.
Current detection techniques typically catch a fraud only after it occurs. This is mostly due to latency that results from the need to connect to AI resources and wait for analysis to come back. This means a transaction attempt with a stolen credit card or identity can often be completed before detection occurs. Telum can thus help companies move from fraud detection to fraud prevention.
The new chip features an innovative centralised design, which allows the full power of the AI processor to be used for AI-specific workloads.
Read more of our latest Industry Updates news stories
