IAG is working with Google Cloud on a way to detect whether photos submitted as part of an insurance claim are genuine or are the work of AI or sourced from the internet.
Executive general manager of data, risk and resilience David Abrahams told a Google Cloud Summit held in Sydney in May that a proof-of-concept for ‘IAG Image Fraud Search’ is currently being worked on.
He presented a tabbed user interface on-stage that appeared to provide a way to upload images and then compare them to results on the web, IAG’s own image repository and to check EXIF metadata.
“One thing that we’re focused on is being able to detect fraudulent images being used in our claims process,” Abrahams said.
“Since the democratisation of generative AI, you can imagine the layer of complexity that this has added to our business processes, like the once-simple process of using photos to assess claims.
“We’re now having to ask questions like, ‘How do we know this image is real?’ ‘How do we know that this image was taken at the point where the accident occurred?’ ‘How do we know that this image hasn’t just been used in another claim or just downloaded off the internet?’
“So, in response to that, we’re developing a proof-of-concept with Google to help us answer all of those questions.”
Abrahams said that the solution reviews images against the stated scenarios and tells IAG “within milliseconds whether these photos are the real deal.”
“This will definitely help us to reduce fraud and costs for our business,” Abrahams said.
The threat of doctored, AI-generated or deepfaked images being used in claims has also similarly seen fellow insurer Allianz invest in technology to detect fraud attempts.