OpenAI Releases Real-Time Video Feature for ChatGPT’s Advanced Voice Mode - 1

Photo by Hc Digital on Unsplash

OpenAI Releases Real-Time Video Feature for ChatGPT’s Advanced Voice Mode

  • Written by Andrea Miliani Former Tech News Expert

OpenAI started to roll out a real-time video feature for ChatGPT’s Advanced Voice Mode which will be available for Team, Pro, and Plus members on the mobile app within the next few days.

In a Rush? Here are the Quick Facts!

  • OpenAI released the real-time video feature for ChatGPT’s Advanced Voice Mode announced in May during the Spring Event
  • ChatGPT Team, Pro, and Plus members will be able to try the update their mobiles in the following days, users in Europe will have to wait longer
  • During the live demo, ChatGPT recognized and remembered OpenAI’s team

The startup shared the announcement on a live streaming session, part of its Christmas program of “12 Days of OpenAI” in which they also launched the $200 model ChatGPT Pro and the video generator tool Sora.

“Just in time for the holidays, video and screen sharing are now starting to roll out in Advanced Voice in the ChatGPT mobile app,” wrote the company on the social media platform X.

Just in time for the holidays, video and screensharing are now starting to roll out in Advanced Voice in the ChatGPT mobile app. pic.twitter.com/HFHX2E33S8 — OpenAI (@OpenAI) December 12, 2024

“We’ll get the feature to Plus & Pro users in the EU, Switzerland, Iceland, Norway, & Liechtenstein as soon as we can. Enterprise & Edu users will have access in early 2025,” the startup added in a comment.

OpenAI shared a preview of this feature in May, during its Spring Even t, surprising the audience with their Chatbot capabilities. Now, ChatGPT gets to “see” through users’ phones and speak as a person would do during a call.

In the presentation, OpenAI’s team introduced themselves to the AI and then ran a test asking ChatGPT if it could remember who was wearing the reindeer antlers and the Santa hat and the AI aced the quiz, interacting with enthusiasm and fluency.

“This is the thing that has felt most like a futuristic UX to me recently,” wrote Sam Altman, OpenAI CEO, on X , in a comment a user made praising the new feature.

To access the new feature, users will need to tap the video icon and allow the “Share Screen” option.

Hackers Exploit Radiant Capital With Malware, $50M Stolen in Heist - 2

Image by Freepik

Hackers Exploit Radiant Capital With Malware, $50M Stolen in Heist

  • Written by Kiara Fabbri Former Tech News Writer
  • Fact-Checked by Justyn Newman Former Lead Cybersecurity Editor

A malware-laced PDF sent to Radiant Capital engineers enabled North Korean hackers to steal over $50 million.

In a Rush? Here are the Quick Facts!

  • Radiant Capital suffered a $50M loss in an October 16, 2024 cyberattack.
  • Attackers impersonated a former contractor, delivering malware through a spoofed Telegram message.
  • Malware enabled malicious transactions while displaying benign data in Radiant’s front-end interfaces.

In a recent follow-up report on the breach, Radiant assisted by Mandiant revealed further details. On September 11, 2024, a Radiant developer received a Telegram message from an impersonated former contractor.

The message, purportedly from a former contractor, included a link to a zipped PDF. Allegedly related to a new smart contract auditing project, the document sought professional feedback.

The domain associated with the ZIP file convincingly mimicked the contractor’s legitimate website, and the request appeared routine in professional circles. Developers frequently exchange PDFs for tasks such as legal reviews or technical audits, reducing initial suspicion.

Trusting the source, the recipient shared the file with colleagues, inadvertently setting the stage for the cyber heist.

Unbeknownst to the Radiant team, the ZIP file contained INLETDRIFT, an advanced macOS malware camouflaged within the “legitimate” document. Once activated, the malware established a persistent backdoor, using a malicious AppleScript.

The malware’s design was sophisticated, displaying a convincing PDF to users while operating stealthily in the background.

Despite Radiant’s rigorous cybersecurity practices—including transaction simulations, payload verification, and adherence to industry-standard operating procedures (SOPs)—the malware successfully infiltrated and compromised multiple developer devices.

The attackers exploited blind signing and spoofed front-end interfaces, displaying benign transaction data to mask malicious activities. As a result, fraudulent transactions were executed without detection.

In preparation for the heist, the attackers staged malicious smart contracts across multiple platforms, including Arbitrum, Binance Smart Chain, Base, and Ethereum. Just three minutes after the theft, they erased traces of their backdoor and browser extensions.

The heist was executed with precision: just three minutes after transferring the stolen funds, the attackers wiped traces of their backdoor and associated browser extensions, further complicating forensic analysis.

Mandiant attributes the attack to UNC4736, also known as AppleJeus or Citrine Sleet, a group linked to North Korea’s Reconnaissance General Bureau (RGB). This incident highlights the vulnerabilities in blind signing and front-end verifications, emphasizing the urgent need for hardware-level solutions to validate transaction payloads.

Radiant is collaborating with U.S. law enforcement, Mandiant, and zeroShadow to freeze stolen assets. The DAO remains committed to supporting recovery efforts and sharing insights to improve industry-wide security standards.