Last year, we published an article about SideWinder, a highly prolific APT group whose primary targets have been military and government entities in Pakistan, Sri Lanka, China, and Nepal. In it, we described activities that had mostly happened in the first half of the year. We tried to draw attention to the group, which was aggressively extending its activities beyond their typical targets, infecting government entities, logistics companies and maritime infrastructures in South and Southeast Asia, the Middle East, and Africa. We also shared further information about SideWinder’s post-exploitation activities and described a new sophisticated implant designed specifically for espionage.

We continued to monitor the group throughout the rest of the year, observing intense activity that included updates to SideWinder’s toolset and the creation of a massive new infrastructure to spread malware and control compromised systems. The target

• X, the former Twitter, has experienced a worldwide outage as of 12PM on Monday CAT.
• This is likely the first major outage of the company since Musk took ownership in 2022.
• The outage seems to have only lasted for about half an hour.

• X, the former Twitter, has experienced a worldwide outage as of 12PM on Monday CAT.
• This is likely the first major outage of the company since Musk took ownership in 2022.
• The outage seems to have only lasted for about half an hour.

Companies are turning to tech solutions to screen candidates. Critics and job seekers have concerns.

Companies are turning to tech solutions to screen candidates. Critics and job seekers have concerns.

• CR researchers were able to easily create a voice clone based on publicly available audio in four of the six products in the test set:

  1. These products did not employ any technical mechanisms to ensure researchers had the speaker’s consent to generate a clone or to limit the cloning to the user’s own voice. These companies—ElevenLabs, Speechify, PlayHT, and Lovo—required only that researchers check a box confirming that they had the legal right to clone the voice or make a similar self-attestation.
  2. Descript and Resemble AI took steps to make it more difficult for customers to misuse their products by creating a non-consensual voice clone.

• Four of the six companies (Speechify, Lovo, PlayHT, and Descript) required only a customer’s name and/or email address to make an account.

• CR researchers were able to easily create a voice clone based on publicly available audio in four of the six products in the test set:

  1. These products did not employ any technical mechanisms to ensure researchers had the speaker’s consent to generate a clone or to limit the cloning to the user’s own voice. These companies—ElevenLabs, Speechify, PlayHT, and Lovo—required only that researchers check a box confirming that they had the legal right to clone the voice or make a similar self-attestation.
  2. Descript and Resemble AI took steps to make it more difficult for customers to misuse their products by creating a non-consensual voice clone.

• Four of the six companies (Speechify, Lovo, PlayHT, and Descript) required only a customer’s name and/or email address to make an account.

It's just one of 6,000 apps that New Zealand thinks might be best tamed with ERP

It's just one of 6,000 apps that New Zealand thinks might be best tamed with ERP

The world's largest contract electronics maker, Foxconn, said Monday it has built its own large language model with reasoning capabilities, developed in-house and trained in just four weeks.

In December 2024, Microsoft Threat Intelligence identified a malware campaign stemming from pirate streaming sites. Using iframe malvertising redirector URLs to generate revenue, and redirects up to five layers deep, malware payloads hosted on GitHub, Discord and Dropbox, acted as a dropper for additional payloads hosted elsewhere. Microsoft says the goal was to steal information and it believes almost a million devices were infected.

Look at you! On your phone! But you’ve got a secret! And you won’t tell! You’re not on your phone! It is only as if you were on your phone! You’re just pretending to be on your phone! On your phone!

It is as if you were on your phone is an almost speculative game about an incredibly near future in which we’re all simultaneously under significant pressure to be on our phones all the time, but also to not be on our phones all the time. Our fingers want to touch the screen, our eyes want to watch the surface, our brains want to be occupied efficiently and always. But it’s also exhausting liking photos, swiping profiles, watching short-form video, and everything else we’re always doing. It is as if you were on your phone presents an alternative: pretend to be on your phone so that you pass as human, but actually do essentially nothing instead. Follow the prompts and be free.

A woman from Dunfermline has spoken of her shock after an Apple voice-to-text service mistakenly inserted a reference to sex - and an apparent insult - into a message left by a garage.

_{The is what Mrs Littlejohn saw on the voicemail screen in the Phone app on her iPhone after receiving a voicemail from the garage.}

A woman from Dunfermline has spoken of her shock after an Apple voice-to-text service mistakenly inserted a reference to sex - and an apparent insult - into a message left by a garage.

_{The is what Mrs Littlejohn saw on the voicemail screen in the Phone app on her iPhone after receiving a voicemail from the garage.}

Large Language Models (LLMs) have demonstrated remarkable performance in solving complex reasoning tasks through mechanisms like Chain-of-Thought (CoT) prompting, which emphasizes verbose, step-by-step reasoning. However, humans typically employ a more efficient strategy: drafting concise intermediate thoughts that capture only essential information. In this work, we propose Chain of Draft (CoD), a novel paradigm inspired by human cognitive processes, where LLMs generate minimalistic yet informative intermediate reasoning outputs while solving tasks. By reducing verbosity and focusing on critical insights, CoD matches or surpasses CoT in accuracy while using as little as only 7.6% of the tokens, significantly reducing cost and latency across various reasoning tasks.

A tale of good versus evil played out on the large screen in the sanctuary of St. Paul’s Lutheran church in Finland. Jesus was shown in robes with long hair and a beard, while Satan was dressed in more modern clothes but with a menacing frown and higher-pitched voice — all created by artificial intelligence.

A tale of good versus evil played out on the large screen in the sanctuary of St. Paul’s Lutheran church in Finland. Jesus was shown in robes with long hair and a beard, while Satan was dressed in more modern clothes but with a menacing frown and higher-pitched voice — all created by artificial intelligence.