Latest AI news - Saturday, May 18 2024

OpenAI's long-term AI risk team, focused on existential dangers of AI, has disbanded following the departures of key researchers, including co-leads Ilya Sutskever and Jan Leike. Their work will be integrated into other research groups within OpenAI. The move follows a broader internal shakeup post-governance crisis.

OpenAI and Reddit have partnered to add Reddit's content to ChatGPT. This move aims to expand Reddit's revenue beyond ads, using its API for AI training. The deal, following one with Google, increased Reddit shares by 12%.

Google is enhancing Android with new security tools in its Android 15 update, including a feature that locks your phone if it's snatched. This theft protection, along with other updates like remote locking and improved AI for detecting threats, will be available for devices running Android 10 or newer.

Google is launching AI Overviews in the US, enhancing search results with AI to make them more useful. Although these new link cards are clicked more often, Google won't share specific data about their performance. This feature will soon be available worldwide, aiming to improve searches with smarter, more organized results.

OpenAI just released a new AI model called GPT-4o, enhancing the previous version to understand and respond to text, speech, and video. This update makes it easier to talk to ChatGPT, making it feel more like a real assistant that understands emotions and nuances in voice. The update also supports multiple languages better and works faster. GPT-4o is now available for all ChatGPT users and will be added to premium services soon.

OpenAI is planning a May 13 event to showcase new updates for ChatGPT and GPT-4. They will also introduce their AI-powered search engine, supported by Microsoft's Bing. This event is timed just before Google's own AI product announcements at Google I/O.