Welcome to Get The Gist, where every weekday we share an easy-to-read summary of the latest and greatest developments in AI—news, innovations, and trends—all delivered in under 5 minutes! ⏱

In today’s edition:

• Google Open-Sources Tool for Watermarking AI-Generated Text

• OpenAI’s New Model Speeds Up AI Image Generation 50x

• Microsoft Photos App Gets AI Capabilities

• Midjourney Launches Web-Based AI Image Editor

• And more AI news….

Top Developments

Google

Google Open-Sources Tool for Watermarking AI-Generated Text

The Gist: Google has launched SynthID Text, a new open-source tool that enables developers to watermark AI-generated text, helping to identify its origins and combat misinformation.

Key Details:

• SynthID Text modifies the probabilities of word selection in AI text to create identifiable watermark patterns.

• The tool still works even after texts are paraphrased, though its effectiveness diminishes with short responses and heavy rewrites.

• While integrated into Google’s Gemini, competitors like OpenAI are exploring their own watermarking solutions, raising questions about future interoperability.

• The potential for a standardized system or legal requirements for watermarking AI content remains uncertain.

OpenAI

OpenAI’s New Model Speeds Up AI Image Generation 50x

The Gist: OpenAI has introduced an innovative continuous-time consistency model (sCM) that drastically speeds up multimedia generation, allowing high-quality images to be created in less than a tenth of a second—50 times faster than traditional methods.

Key Details:

• The sCM achieves impressive sample quality while requiring only two steps, compared to hundreds with traditional diffusion models.

• It operates on a model with 1.5 billion parameters, generating samples in about 0.11 seconds using a single GPU.

• Extensive benchmarking shows sCM delivers top-notch results with significantly less computational effort, making it ideal for real-time applications.

• Its rapid sampling capabilities could revolutionize various fields by enhancing the efficiency of AI-generated images, videos, and audio.

Microsoft

Microsoft Photos App Gets AI Capabilities

The Gist: Microsoft is rolling out a cool new feature called Super Resolution in the Photos app that enhances low-res images using AI, making them up to eight times clearer. Currently exclusive to certain advanced PCs, this update is part of Microsoft’s ongoing AI enhancements for Windows 11.

Key Details:

• Super Resolution increases image quality significantly, perfect for reviving old pictures or enhancing photos for projects.

• This feature is available only on next-gen Copilot+ PCs with a special AI processing chip.

• Currently, only Windows Insiders can access Super Resolution, with no timeline for broad release yet.

• The update also adds useful features like text recognition from images and improvements to the image zoom function.

Quick Gist

Midjourney has launched a powerful web-based AI image editor, allowing users to alter any image, sparking concerns over deepfakes as moderation challenges arise (Read More).

OpenAI appointed Lilian Weng as the new head of its preparedness team to bolster safety measures for advanced AI models amidst growing concerns over AI security (Read More).

Andrej Karpathy launched Eureka Labs, an education-focused AI startup offering an AI teaching assistant and an AI course, LLM101n, designed to help students build AI models (Read More).

OpenAI promoted Michal Pachocki to Chief Scientist, acknowledging his critical role in the company's research and advancements (Read More).

OpenAI's new real-time voice API has raised concerns after researchers demonstrated its potential for automating phone scams, achieving a 36% success rate at a low cost (Read More).

Hugging Face launched Generative AI Services (HUGS), a zero-configuration microservice to simplify the deployment of open-source AI models, offering an OpenAI-compatible API optimized for various hardware accelerators (Read More).

Former OpenAI researcher Suchir Balaji resigned over concerns about the company's use of copyrighted data for AI training, advocating for regulatory reforms due to the impact on the internet ecosystem (Read More).

Samsung's Galaxy AI is expanding language support to 20 languages, including Turkish, Dutch, Swedish, and Romanian, enhancing accessibility for a broader range of users (Read More).

NVIDIA and Reliance announced a partnership to build advanced AI infrastructure in India, positioning the country as a leader in AI services and innovation (Read More).

A mother is suing Character.AI after her 14-year-old son died by suicide, claiming that his attachment to an AI chatbot contributed to his mental distress and ultimate death (Read More).

THAT’S IT FOR TODAY!