FBI Warns AI-Generated Voices Used to Impersonate Senior U.S. Officials in Cyber Campaign

/in /tarafından

The FBI has issued a public warning that malicious actors are leveraging AI-generated voice messages and text to impersonate senior U.S. government officials, targeting both current and former federal and state officials in a sophisticated social engineering campaign.

According to the FBI’s announcement on Thursday, the aim of the scheme is to:

  • Gain access to personal accounts of government officials

  • Target additional contacts once access is gained

  • Harvest sensitive information or even solicit funds fraudulently

How the Scheme Works:

  • Attackers initiate text message conversations to build rapport with the targets.

  • Once trust is established, they urge the recipient to switch to another platform, often linking to a hacker-controlled website designed to harvest credentials like usernames and passwords.

  • In some cases, attackers use AI-generated voice clips to convincingly impersonate the tone and mannerisms of known officials.

Threat Scope:

The FBI has not disclosed how many individuals have been targeted or whether the actors are financially motivated cybercriminals or state-aligned entities. The use of generative AI makes attribution and detection more difficult, and the agency continues to assess the full scope of the threat.

This follows a December 2024 warning from the FBI regarding the broader use of AI-generated content — including text, audio, images, and video — to commit crimes such as fraud, extortion, and identity theft.

Broader Implications:

The campaign underscores the growing threat of generative AI in cybercrime, particularly in impersonation and phishing-style attacks aimed at high-value targets. Government agencies and private sector organizations are now being urged to:

  • Strengthen multi-factor authentication

  • Train personnel to recognize AI-driven impersonation attempts

  • Avoid clicking on unsolicited links or moving conversations to unknown platforms

As AI tools become more accessible, security experts warn that digital impersonation will become an increasingly common tactic for attackers seeking access to sensitive systems or socially engineered pathways into secure environments.

Xiaomi to Launch Self-Developed XringO1 Mobile Chip in Late May

/in /tarafından

Xiaomi is set to unveil its latest in-house mobile chip, the XringO1, in late May, marking a renewed push into proprietary silicon development as competition heats up in China’s smartphone market. The announcement came from CEO Lei Jun on Weibo, though technical details remain limited.

Key Developments:

  • The XringO1 chip was reportedly built using ARM architecture and manufactured by TSMC using its advanced 3-nanometer (3nm) node — currently one of the most cutting-edge fabrication processes.

  • This move signifies Xiaomi’s return to high-end chip design, following earlier efforts like the Pengpai S1 in 2017, which was ultimately discontinued due to cost and complexity.

  • Xiaomi had since shifted focus to image sensors and power management ICs, but resumed processor development in 2021, according to sources familiar with the project.

Strategic Significance:

The launch positions Xiaomi to compete more directly with Apple and Huawei, both of which use custom-designed chips to drive hardware-software integration and performance optimization:

  • Apple has long utilized its own A-series and M-series chips, creating a tightly controlled ecosystem.

  • Huawei, despite U.S. sanctions, has pushed ahead with its Kirin chip series via SMIC.

Xiaomi’s goal is to use the XringO1 chip in its premium smartphones and tablets, potentially phasing in custom silicon to reduce dependency on foreign suppliers such as Qualcomm, which remains Xiaomi’s primary chip partner.

Market & Political Context:

  • The Xiaomi 15 Ultra, launched earlier this year, uses Qualcomm’s Snapdragon 8 Elite, but future models could see a transition to Xiaomi silicon.

  • While U.S. export restrictions now limit TSMC’s ability to produce AI chips for Chinese clients at advanced nodes, smartphone chips — such as XringO1 — are currently exempt from those restrictions.

Xiaomi ranks as the world’s third-largest smartphone manufacturer, and its expansion into EVs, home appliances, and AI devices makes chip autonomy increasingly strategic.

Neither Xiaomi nor Qualcomm responded to requests for comment at the time of reporting.

AI Video Startup Hedra Raises $32M in Series A Led by Andreessen Horowitz

/in /tarafından

Hedra, a fast-growing AI startup specializing in lifelike video generation, has raised $32 million in a Series A funding round led by Andreessen Horowitz’s Infrastructure fund. The round values the company at $200 million, according to a source familiar with the deal, and brings Hedra’s total funding to $43 million since its founding in 2021.

The fresh capital will be used to expand operations, triple its 20-person team, and scale adoption among marketers and creative professionals who are increasingly turning to AI-generated content.

What Makes Hedra Stand Out?

Hedra is entering a competitive space alongside players like OpenAI’s Sora and Runway’s Gen-2, but it aims to differentiate itself through its proprietary Character-3 foundation model — an AI that fuses text, image, and audio inputs to create realistic digital characters.

CEO Michael Lingelbach, a former stage actor, emphasized the company’s goal to overcome the “uncanny valley”, the unsettling effect caused by nearly lifelike avatars. “Getting over the uncanny valley of compelling performance is the hardest frontier in video,” he said, noting that Character-3 is designed to deliver emotionally resonant, human-like performances.

Industry Context:

The AI video sector is witnessing rapid innovation and investment, but concerns remain over:

  • Scalability, due to the high computational costs of rendering realistic avatars

  • The psychological discomfort triggered by avatars that are lifelike but not quite human

  • Ethical questions surrounding the use of synthetic voices and faces in marketing

Backers and Vision:

The round drew returning support from a16z Speedrun, Abstract, and Index Ventures, signaling strong confidence in Hedra’s long-term potential.

As businesses seek cost-effective, scalable ways to create content, Hedra’s technology could play a key role in branding, advertising, and even virtual influencer markets.