AI News - Advancements You Might Have Missed This Week

Meta's AI Ecosystem: App Updates and Privacy Changes

Meta held its first-ever Llamacon event, focusing specifically on AI advancements[cite: 2, 3]. A major highlight was the new version of the Meta AI app, which rebranded the previous Meta View app and introduced a standalone AI chat feature powered by Llama[cite: 4, 6, 7]. The app allows direct conversation with the LLM[cite: 7]. Interestingly, the app aims to make AI chats more social by including a share button that posts conversations to a feed, allowing others to comment, share, and like[cite: 18, 20, 21, 22]. Users can also view other people's creations and chats for inspiration[cite: 23, 24]. The app can also generate images using Meta's Emu AI image generator[cite: 25]. Another convenient feature is the ability to seamlessly continue conversations started on Ray-Ban Meta glasses within the app or web app[cite: 28, 29].

However, recent privacy policy changes for the Ray-Ban Meta glasses have drawn attention[cite: 30]. Meta AI with camera use is now always enabled unless explicitly turned off[cite: 32]. While photos and videos captured on the glasses are stored on the user's phone and not used for Meta's training[cite: 33], voice recordings can no longer be opted out of cloud storage[cite: 34]. These voice transcripts and audio recordings are stored for up to one year and used to improve Meta's products[cite: 35, 36, 37]. There are also plans to roll out ads within the Meta AI app in the future, potentially appearing as interspersed ads in the feed or even sponsored product recommendations within chat responses[cite: 38, 41, 42, 45, 48, 50].

Google's AI Progress: Search, Image Editing, and Language Learning

Google announced several updates this week, including wider availability of its new AI mode in search[cite: 52, 53]. This feature, which provides AI-generated responses with links to sources, is now available for all Labs users in the US and is starting a limited test outside of Labs for a small percentage of US users[cite: 54, 55, 63, 64].

The Gemini app on Android now allows users to upload and edit images directly within the app, a feature previously available in AI Studio[cite: 66, 67, 69]. Notebook LM, a tool for generating audio overviews from data, now supports over 50 languages for these audio summaries[cite: 70].

Google also launched a new app called "Little Language Lessons," designed to help users learn just enough vocabulary and phrases for specific situations, like traveling to a new country[cite: 77, 78]. The app includes features like Tiny Lesson for relevant vocabulary and grammar, Slang Hang for learning expressions from generated conversations, and Word Cam for identifying and learning about surroundings through photos[cite: 79, 80, 81].

Creative AI Tools See Exciting Updates

Several creative AI platforms also rolled out new features. Recraft, an AI image generation and editing platform, introduced a large new style library allowing users to easily apply different artistic styles to their images[cite: 91, 93, 97, 98]. A unique feature is the ability to blend multiple styles and adjust their weighting to create entirely new, customized styles[cite: 112, 118]. These custom styles can also be saved and shared[cite: 115, 116, 119].

MidJourney rolled out "omni reference," a feature that allows users to incorporate a specific character, object, or other element from a reference image into a newly generated image[cite: 194, 195]. This works by dragging and dropping an image into the prompt and adjusting a strength slider[cite: 199, 200, 201].

Cling AI introduced an "instant film effect" that transforms a portrait image into an animated Polaroid-like visual[cite: 208, 209, 210]. Higsfield AI launched "iconic scenes," enabling users to place themselves into legendary movie moments using a selfie[cite: 219, 220].

Craya rolled out "GPT Paint," which allows users to visually prompt the AI for image edits using marks, shapes, notes, and reference images directly on an image[cite: 228, 229]. Users can draw arrows and add text to guide the AI in incorporating elements from other images[cite: 232, 233, 234].

Interestingly, recent observations with GPT-4o image generation show that repeatedly prompting the AI to create an exact replica of an image results in significant changes and drift over multiple iterations[cite: 237, 238, 239, 241, 243].

Other Notable AI Developments

Suno released version 4.5 of their music generation model, featuring minor improvements such as expanded genres, smarter mashups, enhanced voices, and better prompt adherence[cite: 253, 254].

Duolingo announced its intention to become an "AI-first" company, gradually reducing reliance on contractors for tasks that can be handled by AI and focusing employees on creative problem-solving[cite: 256, 257, 260]. Lyft launched an AI earnings assistant to help drivers optimize their shifts based on real-time data[cite: 263, 264, 265]. Aurora has deployed fully autonomous tractor trailers on public highways in Texas for customer deliveries[cite: 266, 267].

Elon Musk also indicated that Grock 3.5 is expected to be released soon for Super GRO subscribers, claiming it will be the first AI capable of accurately answering complex technical questions and reasoning from first principles[cite: 147, 150, 151]. Enthropic announced new integrations for Claude, available on their higher-tier plans, facilitating communication with other apps[cite: 155, 156, 159, 160]. Alibaba unveiled Quinn 3, a hybrid AI reasoning model that allows users to toggle the model's chain-of-thought thinking process[cite: 163, 164, 166]. A new AI agent called VI from Versep was announced, which can interact with a user's computer and applications similarly to a human[cite: 177, 178, 179].

Overall, this week's AI news highlights continued advancements across various domains, from conversational AI and creative tools to autonomous systems and specialized models.

Stay looped into the latest AI news and tutorials by subscribing to the source YouTube channel. [cite: 280, 281] You can also find more AI tools and news at FutureTools.io. [cite: 285, 286] Sign up for their free newsletter to receive updates and gain access to the AI income database. [cite: 287, 288, 289]

Sources:

• AI News: Advancements That Happened This Week