News/Meta
Meta Faces AI Training Ban in Brazil Over Data Misuse Concerns
Meta, the social media giant, has been ordered by Brazilian regulators to stop training its AI models on personal data from Brazilian users, marking a significant setback for the company's AI ambitions in one of its largest markets. Key developments: The Brazilian data protection authority (ANPD) has banned Meta from using public Facebook, Messenger, and Instagram data from Brazil for AI training purposes: The decision follows Meta's privacy policy update in May, which granted the company permission to use Brazilian users' posts, images, and captions for training its AI models. ANPD cited "risks of serious damage and difficulty to users"...
read Jul 2, 2024Meta Tweaks AI Labeling After Mislabeling Edited Photos as Artificial
Meta is updating its approach to labeling AI-generated content after its "Made with AI" tags were confusing users by incorrectly flagging some lightly edited photos as AI-made. Key changes to Meta's AI labeling policy: Meta is tweaking its AI labeling system in response to user feedback and guidance from its Oversight Board: The "Made with AI" label will be changed to "AI info" across Meta's apps, which users can click for more context Meta is working with industry partners to improve its labeling approach so it better aligns with user expectations The problem with the previous labeling system: Meta's AI...
read Jul 2, 2024Meta Is About to Release Its Most Powerful LLM Yet
Meta is set to launch its most powerful Llama 3 AI language model, potentially rivaling OpenAI's ChatGPT-4 while being more efficient and accessible to researchers. Key Takeaways: Meta's upcoming Llama 3 400B model, with over 400 billion parameters, is poised to match the performance of OpenAI's ChatGPT-4 on benchmarks like MMLU while using less than half the parameters, suggesting significant advancements in efficiency. Early testing shows Llama 3 400B scoring 86.1 on the MMLU benchmark, nearly equaling GPT-4's performance with under 50% of the parameters. The improved efficiency could make Llama 3 400B more cost-effective and less resource-intensive compared to...
read Jul 2, 2024Anthropic Considers Partnerships with Google and Meta, Raising Concerns About Industry Consolidation
Anthropic, an AI research company founded by former OpenAI employees, is considering partnerships with tech giants Google and Meta, potentially giving them access to its advanced AI models as they race to catch up to OpenAI's GPT-4. Key aspects of the potential deals: The discussions highlight the complex dynamics in the rapidly evolving AI industry, where collaboration and competition often intertwine: Anthropic is considering licensing its AI models to Google and Meta, which could help these companies accelerate their AI development efforts and narrow the gap with OpenAI. The potential partnerships underscore the value of Anthropic's AI technology, particularly its...
read Jul 1, 2024Character.AI Mulls Deals with Tech Giants Amid Intensifying AI Chatbot Competition
Chatbot startup Character.AI is considering potential deals with tech giants Google and Meta amid increasing competition in the AI chatbot space: Early success and hype: Character.AI had a strong start, attracting millions of users with its engaging, persona-based chatbots that offered a fun and unique conversational AI experience compared to larger rivals. Founded in November 2021 by former Google AI pioneers, the startup quickly gained traction, with users drawn to its hyper-realistic roleplay chatbots mimicking anime characters, TV personalities, and historical figures. Just four months after securing $150 million in funding from Andreessen Horowitz and others at a $1 billion...
read Jul 1, 2024Instagram Influencers Get AI Chatbots: Engaging or Risky?
Say hello to creator-built AI chatbots on Instagram, as the social media platform takes a step towards integrating AI-powered interactions between influencers and their followers, raising questions about the implications and potential risks of this technology. Key development: Instagram creators to get AI chatbots; In a recent interview, Mark Zuckerberg announced that Instagram is testing a new feature allowing creators to build AI chatbots that users can interact with anytime: The AI chatbots will be created in collaboration with selected Instagram influencers using Meta's AI studio. The feature is currently in the testing phase and will be gradually rolled out...
read Jun 28, 2024Zuckerberg Slams Closed AI, Unveils Meta’s AR Glasses Roadmap
Mark Zuckerberg criticized unnamed AI competitors for their closed-source approach, suggesting they seem to think they are "creating God" with their limited access AI products. Open AI as the future: Zuckerberg emphasized his belief in open source AI, arguing that artificial intelligence technology should not be hoarded by a single company to build a central product: He finds it off-putting when tech industry figures talk about building "one true AI", as if they are creating something akin to God. Zuckerberg believes realistically there needs to be many different AIs created to reflect diverse interests, comparing it to how there isn't...
read Jun 28, 2024Meta’s LLM Compiler Is the Latest Breakthrough To Change How Developers Will Code
Meta's LLM Compiler is a groundbreaking suite of open-source models that could revolutionize code optimization and compiler design, making the process faster, more efficient, and cost-effective. AI-powered code optimization: LLM Compiler pushes the boundaries of efficiency by demonstrating remarkable results in code size optimization and disassembly: The model reached 77% of the optimizing potential of an autotuning search in tests, which could significantly reduce compilation times and improve code efficiency across various applications. LLM Compiler achieved a 45% success rate in round-trip disassembly when converting x86_64 and ARM assembly back into LLVM-IR, showcasing its potential for reverse engineering tasks and...
read Jun 26, 2024Meta’s AI Glasses Stumble as Real-Time Translator, but Hint at Promising Future
The Meta Ray-Ban smart glasses struggle to deliver on their promise of a hands-free, real-time AI translation tool for travelers, with mixed results and technical limitations. Key takeaways: A temperamental party trick, not a reliable travel companion; While the glasses could competently translate simple phrases and provide broad summaries, they often failed to parse more complex or colloquial text, and couldn't handle spoken language at all. The AI translation feature only supports a handful of languages (Spanish, Italian, French, and German) and focuses solely on written text, not speech. Instead of providing detailed, word-for-word translations, the glasses tended to paraphrase...
read Jun 25, 2024Apple Rejects Meta AI Partnership, Citing Privacy Concerns; Pursues Other Collaborations
Apple's privacy concerns prevented an AI partnership with Meta Key Takeaways: Apple turned down a potential AI partnership with Facebook parent company Meta due to privacy concerns over Meta's practices: Brief discussions about integrating Meta's large language model (LLM) into iOS took place in March, but Apple decided not to move forward with a more formal partnership. Apple determined that Meta's privacy practices were not stringent enough to meet its standards for safeguarding user data. Ongoing AI Partnerships: While Apple passed on working with Meta, the company has pursued other AI collaborations: Apple signed a deal with OpenAI to integrate...
read Jun 24, 2024Apple and Meta Explore AI Partnership to Enhance iPhone 16, Rival OpenAI
Apple and Meta are reportedly discussing an AI partnership that could significantly enhance the iPhone 16's AI capabilities, potentially helping both companies better compete with OpenAI in the rapidly evolving generative AI market. Key details of the rumored collaboration: The Wall Street Journal report suggests that Apple and Meta are exploring the integration of Meta's generative AI services into iOS 18 and the upcoming iPhone models: Apple has been in talks with major AI players, including Google, Anthropic, and OpenAI, to augment its Apple Intelligence AI features. Plans to include free ChatGPT access and GPT-4o integration in Apple Intelligence were...
read Jun 24, 2024Meta’s AI Mislabels Photos, Sparking Controversy and Raising Questions
Meta's AI mislabeling real photos as AI-generated sparks controversy and raises questions about the reliability and implications of such labels. Key issues with Meta's AI labeling system: The social media giant's "Made by AI" labels are being incorrectly applied to genuine photographs, causing frustration among photographers and users: Several photographers have reported instances where their original photos or those edited using standard tools like Adobe's cropping feature were mistakenly labeled as AI-generated by Meta. Even minor edits using AI-assisted tools like Adobe's Generative Fill seem to trigger the "Made with AI" label, despite the photos being predominantly created by humans....
read Jun 24, 2024Meta AI Dominates ChatGPT and Google Gemini in Comprehensive AI Chatbot Showdown
Meta AI emerges as the strongest AI chatbot, outperforming ChatGPT and Google Gemini in a series of tests spanning emails, recipes, news summaries, math problems, programming, and mock interviews. Methodology and criteria: The three AI chatbots were evaluated based on their ability to provide accurate, thorough, and well-sourced responses across various everyday use cases: For email writing, all three chatbots performed equally well, generating polite, professional, and customizable email templates. When asked for a chili recipe, Meta AI and Gemini provided sources and links, while ChatGPT lifted the recipe from an unknown site without attribution, raising concerns about plagiarism and...
read Jun 21, 2024Meta’s AI Labeling Mislabels Original Photos, Highlighting Challenges of Identifying AI-Generated Content
Meta's 'Made with AI' labeling system faces criticism as photographers report their unaltered images being mistakenly tagged, highlighting the challenges of accurately identifying AI-generated content amidst the rapid proliferation of generative AI tools. Key issues with Meta's AI labeling approach: Meta's automated system for detecting and labeling AI-generated images on its platforms, including Facebook, Instagram, and Threads, has drawn ire from photographers who claim their unaltered photos are being incorrectly tagged as 'Made with AI': Several photographers have shared examples of their original photos, captured with traditional cameras, being labeled as AI-generated, causing confusion and frustration among content creators. Former...
read Jun 21, 2024Facebook & Instagram Memories Fuel AI: Meta’s Controversial Training Data Plans Spark Privacy Debate
Meta is repurposing personal Facebook and Instagram posts as AI training data, raising concerns about the privacy implications of transforming digital memories into fodder for machine learning algorithms. Key points about Meta's AI training plans: Meta recently announced that public posts, photos and even names from Facebook and Instagram will be used to train AI models starting June 26th, effectively treating users' online histories as a time capsule of humanity: Private messages, posts shared only with friends, and Instagram Stories are excluded, but all other public content is fair game for Meta's AI. Europeans have been given a temporary reprieve...
read Jun 20, 2024Meta’s Chameleon AI Bridges Gap Between Open Source and Commercial Multimodal Models
Meta publicly releases open source AI model Chameleon, marking a significant advancement in multimodal AI capabilities that brings open source technology closer to more commercial offerings from Google and OpenAI. Key Takeaways: Chameleon is a new family of AI models from Meta that can understand and generate both images and text, as well as process combinations of the two modalities: The model comes in 7 billion and 34 billion parameter versions, demonstrating strong performance across image captioning, text-only tasks, and non-trivial image generation. Chameleon's fully token-based architecture allows it to reason over images and text jointly, enabling more advanced multimodal...
read Jun 14, 2024Propagandists Exploit AI to Spread Influence, Platforms Must Act
Propagandists Are Using AI Too—and Companies Need to Be Open About It At the end of May, OpenAI reported a significant development in its corporate history - the disclosure that bad actors had misused its products to run influence operations. The company had detected five networks of covert propagandists from Russia, China, Iran, and Israel using OpenAI's generative AI tools to create large volumes of social media content in multiple languages and turn news articles into Facebook posts, all with the apparent intent of improving the quality and quantity of their output. Case in point: OpenAI's report marks an important...
read May 6, 2024Meta’s AI Monetization Plan: Chatbots to Profit
Here is a summary of the key points from the article: Mark Zuckerberg's AI Ambitions: How Meta Plans to Monetize Its Massive AI Investments Meta CEO Mark Zuckerberg has become "more ambitious and optimistic" about the company's ability to lead in AI, and he has outlined three distinct ways the technology can become a "massive business" for the company in the coming years. Business Messaging Zuckerberg envisions Meta's AI moving beyond simple chatbots to become AI "agents" that can handle more complex tasks and solve user problems. He sees revenue from AI-powered business messaging as one of the nearer-term opportunities,...
read Apr 25, 2024Meta’s AI Ambitions: Zuckerberg Outlines 3 Paths to Massive Growth
Mark Zuckerberg Sees 3 Ways AI Will Become a "Massive Business" for Meta Driving the news: Meta CEO Mark Zuckerberg has become "more ambitious and optimistic" about the company's ability to lead in AI. He plans to spend up to $40 billion this year, largely on AI investments. The big picture: Zuckerberg outlined three specific ways he believes AI can become a "massive business" for Meta in the coming years: Business messaging: Developing generative AI tools that allow companies to automate customer interactions and handle more complex tasks. Zuckerberg sees this as a "nearer term opportunity," potentially less than 5...
read Apr 9, 2024Smaller Llama Models Aim to Democratize AI Access
Smaller Versions of Meta's Llama AI Model Coming Soon Meta, the parent company of Facebook, is planning to release smaller versions of its Llama language model ahead of the flagship Llama 3 model launch later this year. This move underscores the growing trend of AI developers offering more cost-effective and lightweight AI model options to the public. Case in point: Meta will launch two small Llama 3 versions this month before releasing the flagship model in the summer. The smaller models, like Llama 2 7B, are faster, more flexible, and cheaper to run than larger language models. Go deeper: Smaller...
read