Meta's Oversight Board identified several weaknesses in rules and processes that pose a risk of unnecessary harm to users, especially women, who are victims of non-consensual explicit deepfakes.
Generative ai RSS
The introduction of features like Generative Shape Fill, Text to Pattern, Style Reference, and Generate Image demonstrates Adobe's commitment to Generative AI features in Creative Cloud apps.
Deezer has also introduced a new AI playlist generator. The feature called 'Playlist with AI' is currently available in beta to a limited number of Deezer subscribers in different countries.
Google is testing a new feature that would allow YouTube Music users to use generative AI features, by entering a prompt to create a radio station with a description, right from the Home feed.
Building on the image generation capabilities of Meta AI, WhatsApp has been spotted working on a feature that lets Meta AI edit and reply to your photos. The feature isn't available to testers yet.
YouTube now lets you report AI-generated content in videos that looks or sounds like you. The video-sharing platform has detailed the steps to file a privacy complaint in an updated support page.
Chinese search engine giant Baidu has launched its new LLM called Ernie 4.0 Turbo. It's an upgraded version of the Ernie 4.0 model, which has existed since last year and was meant to rival GPT-4.
OpenAI has trained a new model called CriticGPT to catch LLM bugs. As the name says, it's meant to analyze ChatGPT's code outputs during the training process and provide critiques for human trainers.
Meta is now allowing users to fill out a short form if they don't want the social media giant to use their personal data to train its existing or upcoming Artificial Intelligence models.
Google has started rolling out Gemini AI side panels to its Workspace apps, including Docs, Gmail, and Drive. The AI chatbot can summarize content and perform different tasks depending on the app.
Anthropic released Claude 3.5 Sonnet, a more powerful AI model that excels in coding, vision, and language understanding. Sonnet outperforms even top-of-the-line models on internal benchmarks.
YouTube said it's testing the ability to share channels via QR codes and summarize live chats during super active live streams using AI. These features are available to small groups of testers.
Google's Gemini app for Android is now rolling out across the UK and Europe. The generative AI chatbot will be available to iOS users as well in the coming weeks through the Google app.
YouTube is experimenting with a new generative AI feature that creates custom images and videos for green screen backgrounds. The feature is currently available to a small group of creators.
Google explained why its AI Overviews feature gave weird answers, like using glue to stick cheese to pizza. It also claimed that many fake screenshots of the feature are floating around.
A future update to Instagram might let you create personalized chat themes using AI. It's also reported that the social media giant is working on the ability to give users early access to features.
OpenAI has signed a multi-year deal with News Corp to access new and archival content from reputed news outlets, including The Street Journal, Business Daily, Barron's, and MarketWatch.
WhatsApp is working on a new feature that allows users to generate custom profile pictures using AI in the instant messaging app. The feature was discovered in WhatsApp Android beta v2.24.11.17.
YouTube has announced its latest AI experiments feature called Dream Track. Creators can use it to generate AI-powered instrumental soundtracks and use them in their YouTube Shorts.
OpenAI could soon expand on its free version of ChatGPT that you can use without an account. Already available on the web, the feature was spotted on the beta version of the ChatGPT app for Android.
The European Union announced today that Microsoft has not sent over requested information on some of its generative AI tools, and has given the company until May 27 to do so, or risk being fined.
Google explained how it narrowed to the Gemini name for its generative AI model. It means 'twins' in Latin, and one of its inspirations is NASA's human spaceflight program that happened in the 1960s.
Not long after OpenAI showcased GPT-4o, Google uploaded a demo video showing a prototype version of Gemini analyzing a camera feed in real time. It's expected to talk more about AI at Google I/O.
Apple is close to finalizing a deal with OpenAI to integrate its ChatGPT into iOS 18. The deal would add generative AI to iPhones, improving features like Siri and other Apple apps.
Given the recent news around AI and the benefits that Apple's latest M4 chip can provide, it comes as no surprise that reports are surfacing that iOS 18 will include AI transcription and summarising.
Apple will be using its own silicon processors to power the upcoming AI features on the iPhone, iPad, and Mac devices. The company will use the M2 Ultra chips in its data centers for the operation.
Meta has announced today that it will be launching new Generative AI tools for advertisers, including image and text generation. It is also expanding the Meta Verified platform to more countries.
Apple is reportedly working on new features for Safari. The browser is expected to get an AI-powered Intelligent Search feature to summarize text, a Web Eraser, and UI enhancements with Safari 18.
Apple is discussing with OpenAI about incorporating AI features on the next version of its iPhone operating system, iOS 18, just a few weeks before Apple's Worldwide Developers Conference (WWDC).
New strings of code in the latest version of Google Maps suggest that the generative AI feature could soon be heading toward Labs, meaning the feature will no longer be exclusive to just Local Guides.
Microsoft and Google, in collaboration with Thorn and All Tech Is Human, have announced new child safety commitments for generative AI features in their various AI software products.
OpenAI has hired its first employee in India, a government relations head, to push for favorable rules and regulations circling AI in anticipation of massive growth opportunities within the country.
A new AI model by Microsoft Research Asia, VASA-1, can create incredibly realistic deepfakes based only on a single photograph and one voice sample. The model easily beats anything available today.
Meta's Oversight Board investigates whether Facebook and Instagram uphold consistent standards of content moderation across the world when users share explicit AI images of female public figures.
The United Kingdom government has begun to draft a set of regulations that aim to govern the most powerful language models such as those behind OpenAI's ChatGPT which aims to limit potential harm.
xAI launched Grok-1.5V, a new multimodal AI that can understand text, images, diagrams, and more. xAI claims Grok-1.5V outperforms competitors on key benchmarks such as RealWorldQA.
Microsoft is introducing a unified Copilot experience in Visual Studio 17.10 that will combine the features of Copilot and Copilot Chat into one package. This version of Visual Studio is out in May.
Google has launched Gemini for Android Studio, a code assistant AI powered by Gemini 1.0 Pro. The company said the feature is free to use but it's language suggests it'll become paid in the future.
A new survey shows that 41% of senior executives at global companies believe there will be a reduction in workforce sizes as they adopt generative AI, contradicting the narrative of AI as a helper.
Google DeepMind and Stanford University found a way how to mitigate the hallucinations and accuracy problems of artificial intelligence-based chatbots. The system that fact-checks AI is using… AI.