Facebook and Instagram Roll Out New Safety Tools for Teens

Meta Platforms is introducing updates to further protect teens on Facebook and Instagram. Starting this week, those under the age of 16 (or under 18 in certain countries) will be defaulted into more stringent private settings when they join Facebook. A similar default was put into effect on Instagram last year. Meta is also restricting “potentially suspicious adults.” For example, adults will be restricted from messaging teens they aren’t connected to and from seeing teens in their People You May Know recommendations. A “suspicious adult” is one that has recently been blocked or reported by a young person. Continue reading Facebook and Instagram Roll Out New Safety Tools for Teens

Google and Meta Are Developing AI Text-to-Video Generators

AI image generators like OpenAI’s DALL-E 2 and Google’s Imagen have been generating a lot of attention recently. Now AI text-to-video generators are edging into the spotlight, with Google debuting Imagen Video on the heels of Meta AI’s Make-A-Video rollout last month. Imagen Video has been used to generate videos of up to 25-minutes at a 24 fps, 1280×768 pixel spec. Imagen Video was trained “on a combination of an internal dataset consisting of 14 million video-text pairs and 60 million image-text pairs,” resulting in some unusual functionality, according to Google Research. Continue reading Google and Meta Are Developing AI Text-to-Video Generators

Nvidia Debuts New AI Model That Quickly Generates Objects

Nvidia Research is introducing a new AI model that largely automates the process of creating virtual worlds, making it easier for developers to populate games and VR experiences with a diverse array of 3D buildings, vehicles, characters and more. Trained using only 2D images, GET3D generates 3D shapes with high-fidelity textures and complex geometric details. GET3D can generate “a virtually unlimited number of 3D shapes based on the data it’s trained on,” according to Nvidia, which says the objects can be used in 3D representations of buildings or the great outdoors, in games or the metaverse. Continue reading Nvidia Debuts New AI Model That Quickly Generates Objects

Instagram Enables Its Users to Upload Longer Video Stories

Photo- and video-sharing social app Instagram is introducing the option for users to upload longer uninterrupted Stories of up to 60 seconds in duration, without automatically converting the video content into 15-second mini-clips as was previously the norm. The Meta Platforms-owned network began testing the update with select users in December 2021 and just announced that the feature is rolling out to all global users. Allowing content creators to share 60-second Stories in one slide is expected to be a welcome change for both users and viewers. Continue reading Instagram Enables Its Users to Upload Longer Video Stories

OpenAI Expands DALL-E 2 Functionality with Facial Uploads

OpenAI has begun allowing users of its DALL-E 2 image-generating system to work with facial image uploads. The program previously allowed only computer-generated faces in an effort to prevent deepfakes and misuse, but OpenAI says improvements to its safety system succeeded in “minimizing the potential of harm” from things like explicit, political or violent content. OpenAI will continue to prohibit use of unauthorized photos and will seek to protect right of publicity, though it remains to be seen how effective that will be. In the past, customers have complained the company was overzealous in its policing. Continue reading OpenAI Expands DALL-E 2 Functionality with Facial Uploads

Meta Developing a Neural Network to Turbocharge Wikipedia

Meta AI, the artificial intelligence unit of Meta Platforms has developed what it believes is a machine learning model that can simultaneously scan hundreds of thousands of Wikipedia citations to check their accuracy. While the Wikimedia Foundation, which runs Wikipedia, already uses bots, Meta’s proposal would be more extensive than anything currently deployed. Trained using a dataset of 4 million Wikipedia citations, the new Meta AI tool analyzes the linked references, verifying corroboration. With more than 17,000 new Wikipedia articles added each month, this is no small feat. Continue reading Meta Developing a Neural Network to Turbocharge Wikipedia

TikTok Stories Can Now Be Shared via Facebook, Instagram

TikTok is launching a new sharing feature that allows TikTok Stories to be published on competing social networks like Facebook and Instagram. The move may increase exposure for TikTok content on Meta Platforms media as the social giant has been taking steps to downgrade recirculated TikTok videos in Reels. Meta recently advised creators it is prioritizing original Reels content on Facebook and Instagram that are programmed to flag third-party watermarks. Piloting since last year, TikTok’s reposting feature recently began rolling out more broadly to TikTok users. Continue reading TikTok Stories Can Now Be Shared via Facebook, Instagram

Stability AI Releases Stable Diffusion Text-to-Image Generator

Stability AI is in the first stage of release of Stable Diffusion, a text-to-image generator similar in functionality to OpenAI’s DALL-E 2, with one important distinction: this open-source newcomer lacks the filters that prevent the earlier system from creating images of public figures or content deemed excessively toxic. Last week the Stable Diffusion code was made available to just over a thousand researchers and the Los Altos-based startup anticipates a public release in the coming weeks. The unfettered unleashing of a powerful imaging system has stirred controversy in the AI community, raising ethical questions. Continue reading Stability AI Releases Stable Diffusion Text-to-Image Generator

Legal Questions Loom as OpenAI Widens Access to DALL-E

OpenAI is expanding its beta outreach for DALL-E 2 by inviting an additional one million waitlisted people to join the AI imaging platform over the coming weeks. DALL-E users will receive 50 credits during their first month of use and 15 credits every subsequent month, with each credit redeemable for an original DALL-E-prompted generation (returning four images) or an edit or variation prompt (which returns three images). Additional credits may be purchased in 115-generation increments for $15. Starting this month, users get rights to commercialize their DALL-E images. However, the move highlights the legal implications of AI and possible copyright infringement. Continue reading Legal Questions Loom as OpenAI Widens Access to DALL-E

Microsoft Pulls AI Analysis Tool Azure Face from Public Use

As part of an overhaul of its AI ethics policies, Microsoft is retiring from the public sphere several AI-powered facial analysis tools, including a controversial algorithm that purports to identify a subject’s emotion from images. Other features Microsoft will excise for new users this week and phase out for existing users within a year include those that claim the ability to identify gender and age. Advocacy groups and academics have expressed concern regarding such facial analysis features, characterizing them as unreliable and invasive as well as subject to bias. Continue reading Microsoft Pulls AI Analysis Tool Azure Face from Public Use

Adobe Debuts ‘Content Credentials’ to Battle Misinformation

Adobe is releasing an open source developer toolkit that aims to prevent the spread of visual misinformation by including additional metadata that Adobe calls Content Credentials. The system is also designed to help content creators indelibly tag authorship to their work. Announced in 2019, the Content Authenticity Initiative (CAI) project has released a whitepaper introducing the system, which is integrated into Adobe software. The CAI has teamed with hardware manufacturers and newsrooms to help ubiquitize its vision. The Associated Press, The New York Times and The Wall Street Journal have signed aboard. Continue reading Adobe Debuts ‘Content Credentials’ to Battle Misinformation

Google’s Imagen AI Model Makes Advances in Text-to-Image

Google has released a research paper on a new text-to-image generator called Imagen, which combines the power of large transformer language models for text with the capabilities of diffusion models in high-fidelity image generation. “Our key discovery is that generic large language models (e.g. T5), pretrained on text-only corpora, are surprisingly effective at encoding text for image synthesis,” the company said. Simultaneously, Google is introducing DrawBench, a benchmark for text-to-image models it says was used to compare Imagen with other recent technologies including VQGAN+CLIP, latent diffusion models, and OpenAI’s DALL-E 2. Continue reading Google’s Imagen AI Model Makes Advances in Text-to-Image

Gen Z Is Drawn to BeReal, a Photo App with No Ads or Filters

French mobile app BeReal, available for iOS and Android smartphones, has become a hit with Gen Z users. Pitched as an authentic alternative to image-driven social apps like Instagram and TikTok, college students and other young adults have driven what Sensor Tower says are nearly 6.8 million downloads in the past two years. BeReal’s primary feature is it pings all users at the same time, once every 24 hours, prompting them to snap and post a photo. Timing of the prompt changes every day in an effort to catch users at random unprepared moments. Continue reading Gen Z Is Drawn to BeReal, a Photo App with No Ads or Filters

DALL-E 2 by OpenAI Creates Images Based on Descriptions

OpenAI has created a new technology that creates and edits images based on written descriptions of the desired result. DALL-E 2, an homage to the surrealist painter Salvador Dalí and the Pixar film “Wall-E,” is still in development but is already producing impressive results with simple instructions like “kittens playing chess” and “astronaut riding a horse.” OpenAI says the tech, “isn’t being directly released to the public” and the hope is “to later make it available for use in third-party apps. “Already some are expressing worry that such a tool has potential to exponentially increase the use of deepfakes. Continue reading DALL-E 2 by OpenAI Creates Images Based on Descriptions

Reddit Is Evaluating Video Reactions to Enhance Discussions

Community forum aggregator Reddit is looking into expanding user-generated video content. Although it hasn’t yet reached the testing phase, reports say the platform is exploring a TikTok-like suite of editing tools, including a reaction feature comparable to Stitches and Duets (which have inspired similar functionality on Instagram’s Reels and Snapchat’s Spotlight). Reddit already supports video, but not reactions. Reportedly, its interest at this early stage is in reactions as an engagement tool for discussions, not in fostering a full-blown creator economy like TikTok, Instagram and Snapchat. Continue reading Reddit Is Evaluating Video Reactions to Enhance Discussions