Neural networks are reportedly helping criminals create cheap virtual fake IDs online

AI in brief A dodgy website is claiming to use AI in creating images of fake IDs that could potentially be used to trick online verification methods.

The site, OnlyFake, is now down, but reportedly touted its abilities to utilize “neural networks” to generate realistic looking photos of made-up driving licenses, passports, and other types of documents for as little as $15. 

In an experiment, 404 Media, an independent tech-focused publication, went through the process of creating a fake Californian ID using a real photo but a false name and address. There was a complete dashboard to enter the desired information and give an electronic signature to make the AI-generated image. 

The resulting photo of the counterfeit card looks very realistic, and could be used to trick websites analyzing people’s identification like banks or social media platforms to create fake accounts to launder money, or, maybe, even possibly, break into other peoples’ real accounts to steal funds.

The image of the sham ID created by 404 Media managed to bypass the online verification methods for the cryptocurrency platform, OKX, for example. 

AI is revealing secrets in scrolls from ancient Rome

A trio of computer science students have won $700,000, the top prize in a competition challenging developers to build AI software to read ancient scrolls burnt to a crisp and preserved from a leftover volcanic eruption thousands of years ago.

The grand prize for the Vesuvius Challenge 2023, funded by former GitHub CEO Nat Friedman and entrepreneur and investor Daniel Gross, was announced this week. First, the contents of a few scripts from the Herculaneum papyri collection were carefully scanned using X-rays generated in a particle accelerator; the effort reportedly cost $70,000, Bloomberg reported.

Next, the digital scans of the 3D scrolls were shared with participants. They had to figure out a way to unfurl the images into flat, readable chunks, detect any signs of writing on them and piece the passages together using AI algorithms. Their results were then given to expert historians to try and decipher.

The winning team managed to uncover four passages of text, containing 140 characters each, where at least 85 percent of the characters were legible. Now, the next challenge is to decipher more words in the four scrolls, and eventually go on towards reading all 800 scrolls in the Herculaneum papyri collection. 

“The primary goal for 2024 is to read 90 percent of the scrolls, and we will issue the 2024 Grand Prize to the first team that is able to do this,” according to the Vesuvius Challenge 2023 blog post. Anyone can enter the competition.

The Herculaneum papyri were recovered in a buried building that is believed to have been where Lucius Calpurnius Piso Caesoninus, Julius Caesar’s father-in-law, once lived.

Top banker trolled in deepfake scam

Mugur Isarescu, the governor of the National Bank of Romania, was targeted in a deepfake scam promoting sham investments in an attempt to swindle victims.

The digitally altered video and audio depicted Isarescu recommending fraudulent financial advice. Since he’s a reputable figure in finance, what he says could mislead people into buying fraudulent stocks and losing money.

“We are extremely concerned about the significant rise of these types of fraud attempts and we urge people to be very careful with every transaction that they make,” said Dan Suciu, a representative of the central bank.

Bloomberg noted that the rise in financial scams in Romania is fueled by recent events like its largest initial public offering and better returns offered by the Bucharest Stock Exchange. 

The deepfake of Isarescu follows hot on the heels of another incident in Hong Kong, where a doctored video of an employee managed to trick the company’s chief financial officer into transferring $25 million into a miscreant’s accounts during a virtual meeting. Police believe AI was used to create the synthetic footage. 

Waymo under investigation for colliding into cyclist

The San Francisco Police Department and California Department of Motor Vehicles are investigating an incident where a driverless Waymo vehicle collided with a cyclist. 

Waymo’s car was stopped at four-way intersection last week, and its view was reportedly blocked by a truck passing by. As the truck turned, the self-driving vehicle started moving forward and crashed into a cyclist that was also turning after the truck, Reuters reported. The car braked, but not in time to avoid running into the cyclist.  

“The cyclist was occluded by the truck and quickly followed behind it, crossing into the Waymo vehicle’s path,” a company spokesperson told The Verge. “When they became fully visible, our vehicle applied heavy braking but was not able to avoid the collision.”

Thankfully, the accident seems like a minor one and the cyclist didn’t need to be taken to hospital afterwards. Officials in California, however, are particularly antsy about self-driving technology amid its ongoing investigation with rival biz Cruise. 

Last year in October, a driverless Cruise car ran over a woman, who was knocked into its path from another incoming vehicle, and dragged her 20 feet along the streets of San Francisco. Cruise’s permit to drive on the roads was temporarily suspended, and it has grounded all of its cars in other states too. 

Meanwhile, federal government agencies, including the US Department of Justice and the Securities Exchange Commission have launched their own investigations into the accident. ®


This website uses cookies. By continuing to use this site, you accept our use of cookies.