These deep generative models were the first to output not only class labels for images but also entire images. Mochi 1 represents a significant advancement in open-source video generation, featuring a 10 billion parameter diffusion model built on our novel Asymmetric Diffusion Transformer (AsymmDiT) architecture. Trained entirely from scratch, it is the largest video generative model ever openly released. Additionally, we are releasing an inference harness that includes an efficient context parallel implementation. Mochi 1 preview is an open state-of-the-art video generation model with high-fidelity motion and strong prompt adherence in preliminary evaluation.

This AI "worm" attack reveals that generative models like ChatGPT have reached capabilities that require heightened security to prevent misuse. Researchers and developers must prioritize safety by baking in controls and risk monitoring before commercial release. Without industry-wide commitments to responsible AI, regulation may be needed to enforce acceptable safeguards across critical domains as systems gain more autonomy. The performance of the Yi models can be attributed to the high-quality data resulting from 01.AI‘s data-engineering efforts.

Runway debuted Act-One, a new feature that generates expressive character performances from a single video and image without motion capture or rigging. genmo ai’s Mochi is an extremely impressive release that showcases how competitive the video generation landscape is about to become — especially with the major dominos (Sora, Midjourney?) still to come. Meta landed its first AI news deal, partnering with Reuters to provide real-time news responses through its AI chatbot across the company’s Facebook, Instagram, WhatsApp, and Messenger platforms. The proportion of internet traffic generated by bots hit its highest level last year, up 2% on the year before, according to the 2024 Imperva Bad Bot Report. Yet without robust privacy and security measures, these agents could present significant risks.

Tencent’s impressive results with fewer active parameters could reshape how we think about scaling systems — potentially offering a more efficient path forward instead of simply making models bigger. Tencent just released Hunyuan-Large, a new open-source language model that combines scale with a Mixture-of-Experts (MoE) architecture to achieve performances on par with rivals like Llama-405B. Apple just started rolling out new developer tools for upcoming Siri screen awareness features with Apple Intelligence, signaling a major enhancement to the digital assistant’s contextual understanding capabilities. Microsoft filed a patent for a ‘response-augmenting system’ designed to combat AI hallucinations, having the model double-check its answers against real-world information before responding to users. Black Forest Labs launched a new upgrade to its FLUX1.1 pro model, featuring a new ‘Ultra’ mode for 4x higher image resolution in text-to-image generations and a ‘raw’ mode for more realistic generations.

NPUs are optimized to process these networks faster and with less power consumption. NPUs, on the other hand, hit the sweet spot by offering a balance between performance and efficiency. This beta release offers an early glimpse into Apple’s long-awaited Siri upgrades and AI integrations that are set to create truly unique and personalized experiences for users. However, as with all beta software, download with caution — as the update might be buggy, and is not the final version of iOS 18 yet. NASA and IBM have joined forces to create Prithvi WxC, a groundbreaking AI foundation model for weather and climate applications — designed to address a wide range of environmental challenges.

But it’s tough to imagine a company tolerating failure rates that high for very long. A report out this month from MIT Technology Review Insights found that 49% of executives believe agents and other forms of advanced AI assistants will lead to efficiency gains or cost savings. This week, genmoai Anthropic released its newest AI model, an upgraded version of Claude 3.5 Sonnet, that can interact with the web and desktop apps by clicking and typing — much like a person. But 3.5 Sonnet with "Computer Use," as Anthropic’s calling it, could be transformative in the workplace.