🚀 Medial Secures Investment on Shark Tank India - Fueling the Future of Professional Social Networking. 🔥

Startup Showcase

Premium Content

Try our Valuation Calculator →

News on Medial

Meta’s new AI image generator was trained on 1.1 billion Instagram and Facebook photos

Arstechnica

Arstechnica · 1y ago

Meta’s new AI image generator was trained on 1.1 billion Instagram and Facebook photos

Meta has launched a standalone AI image generator called "Imagine with Meta AI", based on its Emu image-synthesis model. The AI model was trained using 1.1 billion publicly available Facebook and Instagram images. Previously, this technology was only available in social networking apps. Users can generate images by providing written prompts, and each generation produces four 1280x1280 pixel images. The tool is available in the US and requires a Meta account. However, some limitations were identified in terms of text rendering and handling different media outputs.

Related News

Thanks to AI, the future of programming may involve YELLING IN ALL CAPS

Arstechnica

Arstechnica · 1y ago

Thanks to AI, the future of programming may involve YELLING IN ALL CAPS

OpenAI's DALL-E AI image generator integrated into ChatGPT revealed internal prompts shared between the image generator and the AI assistant, showing that programming in the future may involve human-like communication techniques. The prompts included commands in all-caps for emphasis. OpenAI relied on regular prompt engineering techniques, and being polite to the AI models seemed to yield better results. The large language model GPT-4, which powers the ChatGPT DALL-E interface, was trained on human communication examples scraped from the web, including instances of polite language. The use of all-caps for emphasis in the prompts likely garnered more attention from the AI model.

Instagram will allow users to set video status

Livemint

Livemint · 1y ago

Instagram will allow users to set video status

Instagram is introducing a new feature called video notes, allowing users to add brief video statuses of about 2 seconds. Different from the Stories feature, video notes can only be recorded using the app's front camera and cannot be uploaded from a device's storage. The videos will be visible to close friends and mutual followers for 24 hours. Users can also reply to notes using audio, photos, videos, and stickers. Additionally, Mark Zuckerberg announced potential AI-powered features for Facebook and Instagram, including Emu Video and Emu Edit for creative image and video generation.

Meta’s AI image generator can’t imagine an Asian man with a white woman

The Verge

The Verge · 1y ago

Meta’s AI image generator can’t imagine an Asian man with a white woman

Meta's AI-powered image generator seems to struggle with creating accurate images of Asian and white individuals. The generator often returns images featuring two Asian people, even when specifically prompted for mixed-race or multicultural relationships. It also leans into stereotypes, adding culturally-specific attire without being prompted. The generator tends to represent Asian women as East Asian-looking with light complexions, ignoring the diversity within the Asian community. This indicates biases in the system and highlights the impact of creators, trainers, and the dataset used in AI systems.

Meta’s image generator has a problem with (some) mixed-race couples and friends

Venturebeat

Venturebeat · 1y ago

Meta’s image generator has a problem with (some) mixed-race couples and friends

Meta's AI image generator has been found to reinforce stereotypes and biases when it comes to representing multicultural relationships. Users attempting to create images of mixed-race couples using prompts like "Asian man and white wife" were consistently met with images of two Asians, failing to accurately represent the requested relationship. Additionally, the generator often displayed culturally specific clothing and blended different Asian cultures together. This highlights the limitations and biases within AI systems, which are a reflection of the data and biases of their creators.

Meta’s AI project faces privacy complaints in Europe

Livemint

Livemint · 1y ago

Meta’s AI project faces privacy complaints in Europe

Facebook and Instagram parent Meta Platforms is facing privacy complaints from a European advocacy group over its new AI project. The Vienna-based NGO, Noyb-European Center for Digital Rights, has filed complaints in 11 European countries, urging privacy bodies to stop Meta's project before changes to its privacy policy take effect. The complaints highlight the challenges faced by big tech companies as they use user data to train AI models. Meta maintains that its approach is compliant with privacy laws and aligned with other tech firms. The complaints come as EU lawmakers introduce comprehensive legislation on AI and data privacy.

WhatsApp Beta tests AI feature for unique profile photo generation: Report

Livemint

Livemint · 1y ago

WhatsApp is reportedly working on an AI-based feature that will allow users to create personalized profile photos. The feature, identified in the latest beta update for Android, will let users input a text prompt into the AI image generator, which will then produce a unique profile picture based on the provided details. This new capability aims to enhance personalization within the app and is especially useful for users concerned about privacy. The exact release date for this feature is still unknown as it is currently under development.

An AI video tool just launched, and it’s already copying Disney’s IP

The Verge

The Verge · 29d ago

An AI video tool just launched, and it’s already copying Disney’s IP

AI startup Luma's video tool, Dream Machine, stirred controversy by featuring characters resembling Disney's Monsters, Inc. in its Monster Camp trailer. Questions arose about whether the AI was trained on copyrighted material. CEO Amit Jain clarified it was a user-uploaded image, not independently generated, that led to the likeness. Luma's lack of transparency in how AI is trained and moderated raised concerns about potential plagiarism in AI-generated content. Disney hasn't publicly commented.

Google Photos to add AI detection for enhanced and generated images: Report

Livemint

Livemint · 10m ago

Google Photos to add AI detection for enhanced and generated images: Report

- Google Photos is reportedly working on a feature to identify if an image was generated or enhanced with AI - Version 7.3 of the Google Photos app includes hidden code strings indicating the upcoming functionality - The feature could include tags like "ai_info" and "digital_source_type" to reveal AI-related information and the specific AI tool or model used - This move is part of Google's efforts to combat the spread of misinformation and deepfakes - Details on how the AI attribution will be presented to users are still unclear, but possibilities include embedding data in image metadata or using on-image badges - Google has not announced a specific launch date for the feature yet

Meta’s AI push needs to efficiently deliver a lot more ad growth

Livemint

Livemint · 1y ago

Meta’s AI push needs to efficiently deliver a lot more ad growth

Meta Platforms, the parent company of Facebook and Instagram, is set to report its first-quarter earnings amid high expectations. The company's share price has surged 36% year-to-date and its market cap has surpassed $1.2 trillion. Factors contributing to this growth include a rebound in the advertising business, cost reduction efforts, and the release of its AI language model Llama 3. Analysts predict Meta's Q1 revenue to increase 26% YoY to $36.2 billion, with full-year advertising revenue expected to rise by 17%. However, geopolitical tensions and market shifts could impact ad spending from China and Apple's iOS changes may affect ad growth.

Expect AI-made prescription drugs to be ready for clinical trials in a couple of years, predicts Google DeepMind CEO

Business Insider

Business Insider · 1y ago

Expect AI-made prescription drugs to be ready for clinical trials in a couple of years, predicts Google DeepMind CEO

Google DeepMind CEO Demis Hassabis believes that AI-designed prescription drugs could be entering clinical testing within a couple of years. He stated that they have signed deals with Big Pharma and expect to have AI-designed drugs in clinical trials in the near future. While AI has the potential to accelerate drug discovery and provide personalized care, there is a risk of bias if the models are not trained on diverse data sets. Hassabis also referenced Google DeepMind's recent setback with its Gemini AI image generator, which generated historically inaccurate images, leading to the feature being temporarily taken offline.

Trackers

Active Indian VC’s

OG Capital Email

With a hands-on approach, OG Capital aims to invest in over 20 promising...

Accel Partners Email

Early and growth-stage investments in disruptive technology companies with...

Early-stage venture capital firm investing in technology startups in India. Focus on...

Access All Trackers

Startup Showcase Winners

June 2025

Helping your parents when you are miles away

The Pit Stop Your Cravings Deserve

The next generation E-commerce platform

Enter Ongoing Startup Showcase

Top Users

Trending News on Medial

Download the medial app to read full posts, comements and news.