AI-Powered news roundup: Edition 10
Not a week goes by without dozens of artificial intelligence and generative AI articles making the headlines, and we're here to help. Our bi-weekly AI news roundup will get you up to speed with the most important developments – in less than 5 minutes.
In this edition of the newsletter:
1. Google restores Gemini’s image generation – but with limits
Source: TechCrunch
Google has re-enabled its Gemini chatbot’s ability to generate images of people, a feature paused earlier this year due to issues with historical inaccuracies and racial stereotyping. The update comes with the release of Imagen 3, a new model designed to produce fairer and more accurate images. However, this feature is currently only available to premium Gemini Advanced, Business, and Enterprise users as part of an early access test.
Imagen 3 also improves overall image quality and integrates SynthID, a tool to prevent deepfakes by applying cryptographic watermarks. Alongside this, Google is rolling out Gems, customizable AI "experts," to all Gemini users.
2. Ideogram: A GenAI tool the finally gets words right
Source: TechCrunch
Ideogram AI has unveiled Ideogram 2.0, the latest version of its text-to-image model, aiming to compete with top players in the generative AI space. This release closely follows the introduction of Flux.1 as the primary image generator for Grok on X (formerly Twitter), marking a new era in AI-driven image creation. Competing models in this space now include Auraflow, KwaiKolors, Hunyuan, Lumina, and Kandinsky 3.
Ideogram 2.0 is touted as significantly outperforming other models in key areas like image-text alignment, subjective preference, and text rendering accuracy. The update also brings new features such as an iOS app and a developer API, enhancing its suite of tools.
The model introduces five presets—realism, drawing, 3D, anime, and general purpose—along with a new color palette for more control over image aesthetics. The “Realistic” style offers photorealistic outputs with enhanced textures, while the “Design” preset excels in artistic text generation, ideal for graphic design and social media content.
3. California’s AI bill SB 1047 passes, Governor’s decision awaited
Source: Reuters
California’s SB 1047, a controversial bill aiming to prevent AI-related disasters, has passed the state senate and now awaits Governor Gavin Newsom's approval. The bill targets large AI models, requiring companies that develop them to implement strict safety protocols to prevent misuse, including an "emergency stop" button and third-party audits. It applies to models costing over $100 million to develop, potentially impacting tech giants like OpenAI, Google, and Microsoft.
领英推荐
While supporters argue the bill is a proactive step to avoid catastrophic AI misuse, it faces strong opposition from Silicon Valley, including venture capitalists and tech companies who claim it could stifle innovation and burden startups. Prominent figures in AI, such as Geoffrey Hinton and Yoshua Bengio, support the bill, citing the need for safeguards against potential doomsday scenarios.
In a surprising move, Elon Musk, who has long advocated for AI regulation, publicly supported the bill, stating on X (formerly Twitter), “This is a tough call... but I think California should probably pass the SB 1047 AI safety bill.” Musk’s own AI company, xAI, would be affected by the bill, despite his plans to leave California.
Meanwhile, OpenAI has opposed the bill, advocating for a federal alternative instead. Other opponents, including major AI researchers and industry leaders, argue that the bill could harm California's tech ecosystem and should be regulated at the federal level instead. The bill’s fate now rests with Governor Newsom, who must decide whether to sign it into law by the end of August.
4. Lumalabs challenges Sora rival Dream Machine 1.5
Source: Lumalabs
Luma Labs has unveiled Dream Machine, a cutting-edge AI model that generates high-quality, realistic videos from simple text descriptions. Dream Machine is designed to create 5-second video clips with fluid motion and cinematography, producing 120 frames in just 120 seconds. This model excels at understanding interactions between people, animals, and objects, ensuring character consistency and accurate physics in its video outputs.
One of Dream Machine’s standout features is its ability to simulate cinematic camera movements, like Dolly Shots and Crane Shots, making it a powerful tool for directors and filmmakers looking to streamline the shot planning process.
Dream Machine is available for free on the Luma Labs website, but high demand has resulted in long wait times, with some users reporting delays of several hours. The free plan includes 30 video generations per month, while paid plans offer up to 2,000 generations.
Explore Dream Machine at https://lumalabs.ai/dream-machine .
5. Microsoft delays Recall AI feature for Windows testers until October
Source: The Verge
Microsoft has postponed the release of its controversial Recall AI feature for Windows testers until October, citing the need for further security testing. Originally set to launch in June with Copilot Plus PCs, Recall has faced delays due to concerns about its security. The feature uses local AI models within Windows 11 to capture and save screenshots of nearly everything on your PC, allowing users to search and retrieve past activities through an interactive timeline.
Despite Microsoft’s assurances of Recall's security and privacy, researchers discovered that the database storing these screenshots wasn’t encrypted, raising the risk of unauthorized access. In response, Microsoft is now making Recall an opt-in feature, encrypting the database, and adding authentication through Windows Hello.
The delay suggests that Recall may not see a full launch this year, as Microsoft typically tests features with Windows Insiders for several weeks or months before a broader rollout. The company has emphasized that security remains the top priority and will provide more details in an upcoming blog post when the feature becomes available for testing in October.