Marco Rubio’s Deepfake sounds emphasizes the risks of AI cyber security

An audio depth mimiced his voice with a member of the Congress with a member of the Marco Rubio, Foreign Ministers, US Governor and AI, according to A high-level official of the United States and a state department dated July 3 cable.

Message recipients were made to be deceived by a scam, designed to make sensitive information or receive account access. But the incident is the high-level example of how easy and exciting convincing-AI sound scammers.

With the speech of someone who is up to the services such as eleven laboratories, speaking and responders, you can only eighty 15-30 seconds, write any message and read aloud in their voices. Remember, these are perfectly used tools for many things for creating content – but as many AI technologies, it can be abused by evil actors.

Deepfakes Danger has increased

AI generated depths are new, especially C-Suite leaders and government officials, but it becomes a bigger problem. Eight months ago, I declare The investigation of more than half of the General Information Security Workers (CISOS) ranked videos and audio depths as an increased concern. This threat has only increased. A new study by Surfshark, in the first half of 2025, in the first half of 2024 in the 580s and in the 580s in the 580s and in 2024 (150 events) and 64 in 64 events. SCHEM: Makes public figures to promote fake investments that result in a loss of $ 401 million.

“The depths have become realistic, active cybertore threats,” AvIAD Mizrachi, CTO and Software Security Company Frontegg, informed me by e-mail. “We already see the created video calls that the employees are successfully monitoring the powers of many dollar payments. These attacks are now and are exciting for a hacker.”

Part of the problem, mizrachi, traditional authentication methods-usernames, passwords, disposable codes and identifier applications are designed for a world where a world can clone your voice or face. Because these cheats are not necessarily involved in entering an account, they trust a real person to rely on to hand over or allow themselves to allow themselves.

“Traditional security measures are no longer working to examine the identity of a person,” he said, “he said most of the most depths deposits in cyberscent groups, and it is the maintenance of sensitivity aggressions. In a voice mail message or video call, convincing fake voice can confirm a wire transfer, even if all the identification means is technically.

To protect this kind of deceit, organizations should place strong safety vehicles that rely on a smartphone or hardware security key to physical devices to prove someone’s identity. These tools are known as Fido2 or Webauthn Passkez, the hackers are more difficult until fake or trick. In addition to the device inspection, smart verification systems are also suitable for anomalies where the speed, location or access habits, clonon sounds like behavioral signals. These additional layers become more difficult to succeed for a depth attack.

Security and EU Strategy Director Margaret Cunningham Darktrace, Rubion’s strange attempt, said that only a generative AI’s convincing, target engineering attacks can be used.

“This danger is poor, because this failed for missing the correct moment of human sensitivity,” he said. “People often do not decide in the center of calm, in the spotlight. Multi, under pressure and managing the acquaintances, can be warned in those moments, a reliable voice or formal point of view.”

Deephzakes affected democracies in the world

The Generative AI, when it comes to the media manipulation, reduced a stronger, cheaper and more expandable, and reduced the barrier that can be expanded. And this affects the world of democracies: recently New York Times report AI-powered depths found that the elections in at least 50 countries in the world have changed the opposition in at least 50 countries using the opposition.

“This is a new border for impact operations,” said Leah Siskind, the EU research officer in the Democracy Fund. ” This is an emergency national security problem with serious diplomatic arrangements. ”

Still, SiSind, government officials recommend cleaning from popular encrypted platforms such as a signal to content, and there are no identity verification mechanisms. “Given the convenience of creating real-looking accounts in any consumer class messaging application, the high state officials should stick to safe communication channels,” he said.

Note: See this new Fortune your video IBM’s quantum computing test lab is about tour. I am an incredible time hanged in IBM in Yorktown Heights Campus (a midcentury modern miracle, St. Luis Arch and JFK Airport in New York). Video was part of the coverage of coverage for this year Fortune 500 issues covering an article in the last ribaverag of IBM.

As I said, “The IBM is like the research center, which is once developed in two worlds, with the original Eeames, the most developed ones in the 20th century modernization, the world’s cooperation and work laboratory, the field of cooperation and work laboratory, the surrounding area of ​​cooperation, the surrounding area of ​​cooperation, the surrounding area of ​​cooperation, the surrounding area of ​​cooperation, the surrounding area of ​​cooperation and work laboratories, the surrounding area of ​​cooperation and work laboratories.

This is the rest of the AI ​​news here.

Sharon Goldman
Sharon.Goldman@fortune.com
@SharongoldMan

In AI News

Xai releases Grok 4 with the return of anti-Semit posts. Elon Musk’s Xai has initiated Grock 4, a few months after the previous release, stressed the rapid pace of the AI ​​development. Late late night, appeared during the Livestream, Musk, New Chatbot allegedly preferred most graduates along the subjects and now supported the voice interaction. Benchmark results that defeat their opponents like Xai, Grock 4 Openai. Musk bot received that the bot sometimes does not have a common sense and has not yet made scientific progress, “This is only one time.” The release of the Grock 4 comes after the Xai’s Xai’s Xai’s Chatbot is a day after facing an anti-Semitic content.

Portellexity launches AI browser from starting from power users on Google. Increases increase Google Avenue with the start CometThe first AI-powered Web browser. Initially, there is a total of 200 $ / monthly subscribers and a limited waiting list, Comet, Confusion Signature Combines the generalized answers instead of the front and traditional search results of the AI ​​search engine. A browser AI is also debuting a browser AI EU agent of the AI ​​agent, which can summarize emails and calendar events and interact with webpages in real time. CEO Aravind Srinivas is more than just a browser than a browser, a step that is a step that is called the “AI Operating System” designed to deepen the daily workflows. The movement, chromium and Google’s own AI search experiences are becoming more directive in the browser, starting betting bets, how people find online data and find online information and interact on the Internet.

Openai has been preparing to launch its own web browser. According to ReutersOpenai also joins the browser wars, as well as to compete with Google, and now confused. According to the report, the report reports that the report is designed to be able to debut in the coming weeks in the coming weeks and to use some user interfaces to use some user interactions to foreign websites. The action can also provide Openai to the user information that lasts Google’s advantages for a long time.

We have fortune

Apple’s AI efforts ‘hit half the night’ and the only way to stop behind the analytics says ives “Marko Quiroz-Guitierrez.”

Would you replace CEO with AI Avatar? -By Alexandra Sternlicht

Why prioritize COCA-COO’s great impressive AI pilot projects – John Kell

Amazon’s tariff-cloudy, seller-mixed, EU explored, the most strangest day -Well Jason Del Ray

There is a calendar

July 13-19: International conference side Machine learning (ICML), Vancouver

July 22-23: Fortune Brainstorm AI Singapore. Apply to attend here.

July 26-28: World Artificial Intelligence Conference (WAIC), Shanghai.

September 8-10: Fortune Brainstorm Tech, Park City, Utah. Apply to attend here.

October 6-10: World AI Week, Amsterdam

December 2-7: Neurips, San Diego

December 8-9: Fortune Brantortorm AI San Francisco. Apply to attend here.

Eye in AI numbers

80%

According to a new report of Gartner, how many work programs will be engaged in the work program, including images, videos, audio and other information, including more work programs, including photos, videos, audio and other information. A large jumps of less than 10% today.

This turn, known as “Multimodal AI”, can change how businesses work in industry as health, finance and production. For example, analyzing the mixture of the program (such as medical pictures) analysis can help make the program smarter decisions and Patient notes) or even active steps to attract the type of art of throwing fraud or supply chains – to attract without human access.

Gartner analysts say that the companies will begin to invest in this new AI technology to be competitive on new AI technologies and to make real value to their customers.