State-of-the-art audio deepfake detectors are no match for well-equipped bad actors, according to a recently published research study.

A trio of researchers from Hamad Bin Khalifa University have discovered that some AI-based audio authentication systems are vulnerable to adversarial attacks, a type of attack that deceives a machine learning model by manipulating its input data.

The trio, which comprises Roberto Di Pietro, Spiridon Bakiras, and Mouna Rabhi, was able to unearth the finding by engineering a set of attacks on Deep4SNet, an audio deepfake detection model with a 98.5% accuracy rate in classifying fake speech. Deep4SNet detects fake audio by converting inputted audio data into histogram images. The model, which is “highly accurate” in image classification tasks, then uses a convolutional neural network that has been trained with more than 2,000 original and fake voice recordings to classify if the audio sample is real or fake.

Fool me twice. During the study, the research team focused on their ability to perform a graybox attack, where the attacker has limited knowledge of the victim’s model, against Deep4SNet. In this case, the attacker would have access to the data used to train Deep4SNet.

Read the rest here.—BM

Broadcom has fired a broadside at AT&T, claiming in court filings that the telecommunications giant’s ongoing lawsuit over VMware licenses is a distraction from its true intent—ditching VMware entirely.

Broadcom closed a deal to purchase VMware for $69 billion in 2023, giving it control of the world’s most ubiquitous virtualization software products. The acquisition immediately sparked concern among VMware customers that Broadcom would aggressively bleed its new holdings for all it was worth—a recent survey of 300 IT decision-makers found around 75% expected VMware costs to at least double. Broadcom’s decision to eliminate perpetual licensing in favor of a subscription model was a particular sore point.

AT&T’s dispute with Broadcom is focused on what obligations, if any, VMware’s new owner has to the owners of those perpetual licenses. In early September, The Register reported that AT&T (which has perpetual licenses) had filed a lawsuit claiming Broadcom refused to honor contract provisions that give it the option to extend support services for another two years. AT&T further claimed Broadcom would only honor the contract if AT&T “agrees to purchase scores of subscription services and software” that it “does not want or need.”

Read more here.—TM

Timing is everything, and as the C-suite found in September, that matters for coming or going.

Here’s who’s in and out for the month.

Mira Murati resigns from OpenAI

After six and a half years, OpenAI CTO Mira Murati made the sudden announcement that she would be leaving the company. In a September 25 message to staff that she also shared to her X account, Murati said that she was leaving OpenAI to do her “own exploration.”

“There’s never an ideal time to step away from a place one cherishes, yet this moment feels right,” Murati wrote.

It marks the end of an era. After working as an engineer with electric car company Tesla, Murati joined OpenAI in 2018, helping to transform it into the behemoth it is today. In a statement posted to X, CEO Sam Altman thanked Murati for her work.

“I feel tremendous gratitude towards her for what she has helped us build and accomplish, but I most of all feel personal gratitude towards her for the support and love during all the hard times,” Altman said. “I am excited for what she’ll do next.”

Keep reading here.—EH

Bring your team and save big! IT Brew is offering 50% off tickets to our cybersecurity event in New York on Oct. 31 for groups of three or more. Whether attending in person or via livestream, this is the perfect opportunity for your team to gain insights from top industry leaders. Use code: GROUP to secure your discounted passes today and stay ahead of the latest cyberdefense strategies!