Short

Elon Musk's AI company, xAI, has just released its Grok chatbot as a free iOS app for US users. The app gives people access to xAI's latest language model, Grok 2, and lets them ask questions, create images, and analyze photos they upload. To keep its answers current, Grok pulls information from Twitter and the web. For now, the app is limited to iPhone users in the US. Musk had previously announced that Grok 3 would arrive by the end of 2024, calling it the "the world's most powerful AI by every metric by December," but it hasn't been released. If the rumor mill is right, Grok 3 is currently being tested and could be released in the coming weeks.

Image: xAI via App Store
Short

An engineer who goes by "STS 3D" caught OpenAI's attention - and not in a good way. The developer built a rifle system that could be controlled through ChatGPT voice commands, demonstrating in a viral video how the system could automatically aim and fire at targets using OpenAI's Realtime API. OpenAI quickly stepped in, revoking the developer's API access for violating their terms of service. While OpenAI has its own military contracts, the company has a strict policy against anyone using tools like ChatGPT for weapons systems.

Image: via Instagram
Ad
Short

Google is experimenting with a new feature called "Daily Listen" that uses AI to create personalized podcast-style summaries in its Discover feed. The system pulls together a roughly five-minute audio episode based on topics and stories you follow, serving it up in the carousel below the search bar in Google's mobile apps. If you're in the US and want to try it out, you can enable Daily Listen through Search Labs on either Android or iOS. Once activated, your first AI-generated episode will show up the following day.

The technology seems to share DNA with Google's NotebookLM podcast system. Like other AI tools that process text, these audio summaries make up facts, especially when trying to digest and synthesize large amounts of diverse content. It is better to treat these summaries as helpful but fallible assistants than as authoritative sources.

Simon via X
Ad
Ad
Short

Las Vegas police revealed that Matthew Livelsberger, a 37-year-old former soldier, used ChatGPT and other AI tools to plan an attack involving a Tesla Cybertruck outside the Trump Hotel on New Year's Day. After detonating the vehicle, Livelsberger died by suicide at the scene. According to Sheriff Kevin McMahill, Livelsberger utilized generative AI to research explosives, projectile velocities, and fireworks before carrying out the attack. McMahill called the use of AI in attack planning a "game changer" and has alerted other law enforcement agencies about this development. OpenAI pointed out that ChatGPT only provided information that was already available to the public. The explosion resulted in minor injuries to seven people, with minimal damage to the hotel structure. Investigators determined that Livelsberger acted alone and had not intended to harm others in the attack.

Google News