Sometimes artificial intelligence should be significantly upgraded
Mail Online, February 11, 2022
An American allegedly used the government’s facial recognition system to claim nearly a million dollars in fake unemployment benefits from a state over 2,000 miles away. All he needed was a wig.
Eric Jaklitsch was arrested late last year after the FBI, DHS and the California Department of Employment summed up the results of an investigation that took almost a year. The man was found to be involved in 78 scams and unemployment claims at The Golden State. Federal prosecutors commented that the scammer attempted to steal more than $2.5 million from public funds. In doing so, he successfully stole approximately $900,000 after he scammed software 0, all he had to do was wear a big curly orange wig to check for fake driver’s licenses from several states.
‘During the scheme, Jaklitsch collected personal identifying information of numerous individuals – including names, birth dates, and Social Security numbers – and used their identities to file fraudulent unemployment insurance claims,’ US Attorney Phillip Talbert’s office explained. Such a simple disguise was enough to repeatedly deceive the state unemployment department, the Employment Development Department (EDD).
Facial recognition systems have once again demonstrated their imperfection and the fact that they need further refinement now in order to avoid the possibility of serious offenses by fraudsters.
Business Insider, February 6, 2022
Authorities are investigating Amazon for the sale of a food preservative allegedly used to aid suicides.
Amazon CEO Andy Jassi has been asked to comment on the platform’s sale of a hazardous chemical compound. In particular, the legislators asked for details about suicides that were previously reported to have been caused by the product. According to a number of sources, 10 people have committed suicide using the compound after purchasing it from Amazon. The whole situation was further complicated by the fact that, since enough people bought the preservative to try to commit suicide, according to the algorithm used by the marketplace, the site’s smart system began to recommend other products to buyers, which are often bought in tandem with a suicide chemical.
“[The chemical] is commonly used to preserve foods such as meats and fish, and can also be used in laboratories as a reagent,” the Amazon spokesperson commented. “It is a widely available product offered by retailers, and unfortunately, like many products, it can be misused.”
Wired, February 11, 2022
Just as important as fighting fraud is, privacy and anti-bias issues should not be neglected.
According to the Treasury Department’s oversight body, the IRS needs to pay attention to protecting against identity fraud. According to the report, as a result of recent data breaches, “much of the information the IRS uses to verify the identity of taxpayers may have been stolen.”
The importance of this issue was also emphasized by the pandemic. The IRS launched a web page where taxpayers could enter bank details for a range of transactions. At the same time, they entered data such as the date of birth of the person and the social security number. Some people logged in to find that the scammers got there first.
In response to this situation, the IRS has introduced a new technology to verify people online based on facial recognition. That said, last year it was awarded an $86 million contract with ID.me, which verifies a person’s identity by using algorithms to match an ID image with a photo from a video or selfie. In November, the IRS launched ID.me for several online services.
However, in early 2022, the IRS’ new security system became a political issue – taxpayers, activists and lawmakers complained that facial recognition could be biased. The IRS responded by saying that it would “refuse to use a third-party facial recognition service,” while offering no alternative way to protect taxpayer data.
The conflicting requirements for the IRS illustrate a longstanding problem in the online age. On the Internet, no one knows who you really are, and it is very difficult to prove that you are exactly who you say you are.
Stay up to date with what is happening! Get a first look at news, noteworthy research and worst attacks on AI delivered right in your inbox.
Carrying out attacks on machine learning models as part of the study is necessary for further successful work on potential vulnerabilities. And here is a selection of the most interesting ...
