Category: Artificial Intelligence

• AI image generator hands are a standout feature of FLUX, a new suite of text-to-image models launched by Black Forest Labs, which aims to improve generative AI for images and videos.
• Founded by former Stability AI engineers, Black Forest Labs offers three versions of FLUX.1, claiming superior performance compared to existing models like Midjourney and DALL-E.
• The FLUX.1 models utilize a hybrid architecture with 12 billion parameters, enhancing their ability to generate accurate images, including human hands, previously a weakness in earlier models.
• The company secured $31 million in funding from notable investors and plans to leverage FLUX.1 for future video generation projects, competing with other advanced models.
• Concerns arise regarding the ethical implications of the training data used for FLUX.1, as it may involve unauthorized image scraping, similar to issues faced by Stability AI.

https://arstechnica.com/information-technology/2024/08/flux-this-new-ai-image-generator-is-eerily-good-at-creating-human-hands/…

• Apple is working to enable Apple Intelligence in China and the EU, facing regulatory challenges that currently block AI features in these regions.
• At WWDC 2024, Apple announced that Apple Intelligence would initially support only US English, with plans to add more languages over the next year.
• CEO Tim Cook emphasized the company’s engagement with regulators to understand requirements before committing to a timeline for feature availability in China and the EU.
• Users can preview Apple Intelligence by switching their devices to English and setting the region to the United States, but access is entirely restricted in China and the EU.
• The AI features will be free with iOS 18 and macOS 15 updates, but will only be available on the latest Apple devices, such as M1 or later iPads and Macs, or the iPhone 15 Pro.

https://9to5mac.com/2024/08/01/apple-intelligence-china-eu/…

• Apple AI investment comparison reveals that while Apple is increasing its spending on AI, it still lags significantly behind competitors like Microsoft, Google, and Meta in capital expenditures for AI infrastructure.
• CEO Tim Cook acknowledged a rise in costs related to AI, reporting $2.15 billion in capital expenditures, which is a modest increase compared to its peers’ substantial investments.
• Unlike its rivals, Apple does not have a cloud business model focused on renting out infrastructure, opting instead for a hybrid approach that shares costs with partners.
• The company is currently testing its Apple Intelligence features, which aim to enhance Siri and automate various tasks, but these are only available to developers at this stage.
• Apple’s integration of OpenAI’s ChatGPT technology into iOS later this year may present future monetization opportunities, despite the confidentiality of their agreement.

https://www.cnbc.com/2024/08/01/apple-spending-more-on-ai-but-remains-behind-its-silicon-valley-peers.html…

• Apple is increasing its spending to prepare for the Apple Intelligence launch this fall, focusing on AI development and resource reallocation.
• The company’s third fiscal quarter of 2024 reported $85.78 billion in revenue, marking a 5% growth in sales.
• CEO Tim Cook expressed optimism about the potential impact of AI on future sales, despite uncertainties.
• Apple Intelligence features will be available for free with iOS 18 and macOS 15 updates, but require the latest devices like M1 or later models.
• The upcoming iPhone 16 models will also support Apple Intelligence, potentially driving new phone sales.

https://9to5mac.com/2024/08/01/tim-cook-spending-more-apple-intelligence/…

• Microsoft Places OpenAI on its list of AI/search rivals, alongside competitors like Amazon and Google, as noted in its fiscal 2024 report.
• The relationship between Microsoft and OpenAI has been crucial, with Microsoft providing $13 billion in funding to support OpenAI’s AI system development.
• Both companies are competing for a larger share of the AI market, particularly in software development and customer service applications.
• OpenAI has entered the search arena with its new tool, SearchGPT, which aims to compete with Microsoft and Google in enhancing search capabilities.
• Advanced natural language processing in SearchGPT allows it to understand user intent better, offering tailored responses to queries.

https://www.pymnts.com/artificial-intelligence-2/2024/microsoft-places-openai-on-list-of-ai-search-rivals/…

• Google makes it easier to remove AI-generated deepfakes content from Search, specifically targeting explicit nonconsensual images.
• The update aims to filter explicit searches related to individuals when removal requests are made, effectively lowering the visibility of such content.
• Google will also remove duplicates of requested explicit images from Search results, although this doesn’t eliminate them from the internet entirely.
• The company is enhancing its ranking system to prioritize high-quality nonexplicit content, aiming to reduce exposure to explicit results by 70%.
• Requests for removal must meet specific criteria, and Google will review each one, potentially seeking additional information before responding.

https://www.cnet.com/tech/services-and-software/google-makes-it-easier-to-remove-explicit-ai-generated-deepfakes-from-search/…

• Gemini AI model benchmarks show that Google’s new Gemini 1.5 Pro has surpassed OpenAI’s ChatGPT-4o and Anthropic’s Claude-3, taking the top slot on the Chatbot Arena leaderboard.
• The experimental release of Gemini 1.5 Pro on August 1 scored 1,300, outperforming previous versions and its competitors, which received scores of 1,286 and 1,271 respectively.
• This marks a significant shift in the generative AI landscape, as OpenAI’s models had dominated benchmarks for the past year with little competition.
• Users have expressed excitement about Gemini’s performance on social media, with some claiming it “blows 4o out of the water.”
• The future status of the experimental version remains uncertain, as it may undergo changes or be rescinded for safety or alignment reasons.

https://cointelegraph.com/news/google-gemini-ai-model-dominates-benchmarks-beats-gpt-4o-claude-3…

• Microsoft OpenAI competitor AI Content highlights that Microsoft has now classified OpenAI as a competitor in its SEC filing, despite a long-term partnership involving a $13 billion investment.
• This classification places OpenAI alongside other competitors in AI and search, including Anthropic, Amazon, Meta, and Google.
• Microsoft is likely attempting to shift the narrative on its relationship with OpenAI amid ongoing antitrust investigations by the FTC.
• The tech giant has distanced itself from OpenAI by relinquishing its board observer seat and investing in its own AI initiatives, like Microsoft Copilot.
• Historical parallels are drawn to past tech partnerships, suggesting that Microsoft and OpenAI’s relationship could evolve similarly as competition intensifies.

https://techcrunch.com/2024/08/01/microsoft-now-lists-openai-as-a-competitor-in-ai-and-search/…

• Google Photos AI editing tools now available for free on Android and iOS
• Initially limited to Pixel phones and Google One subscription
• Includes Magic Eraser to remove unwanted elements, Magic Editor to edit images, and Photo Unblur tool
• Tips shared for maximizing AI tools, such as using Portrait preset and understanding differences between Erase tool and Magic Eraser
• Magic Editor offers three selection methods, and editing tools provide strength slider for varying effects

https://www.zdnet.com/article/you-can-now-use-google-photos-ai-editing-tools-on-android-and-ios-for-free/…

• Nothing introduces AI-powered ‘News Reporter’ widget for Nothing and CMF smartphones
• Widget provides curated news summaries in categories like entertainment, technology, science, and more
• AI system compiles articles from English language publishers and converts them into audio summaries
• Each story summary lasts one minute, with a maximum of eight summaries per day
• Available on Nothing OS 2.6, no sign-up required, and can be controlled through Quick Settings

https://9to5google.com/2024/07/31/nothing-news-reporter-widget-ai/…

• OpenAI ChatGPT introduces advanced voice mode after safety delays, addressing concerns from previous incidents
• ChatGPT Plus subscribers are the first to experience the revamped voice functionality, offering interactive and responsive conversations
• OpenAI aims for dynamic engagement similar to the GPT-4 launch demo, allowing users to interrupt and request changes seamlessly
• Safety measures implemented include stress-testing with external experts, content filters, and collaboration with voice actors
• Limited rollout currently, with plans to expand access to all ChatGPT Plus users in the fall, emphasizing responsible AI development

https://www.businesstoday.in/technology/news/story/openai-begins-rolling-out-advanced-her-like-chatgpt-voice-mode-after-safety-delays-439521-2024-07-31…

• Apple chose Google’s TPU processors over Nvidia hardware for AI training
• Apple’s research paper delves into the development of language models for AI
• The decision to use Google’s TPUs may be influenced by Apple’s strained relationship with Nvidia
• Apple’s AFMs were trained on 8,192 TPU v4 chips and 2,048 TPU v5 processors
• Apple’s AFMs outperformed some competitor models in human evaluations, particularly in generating safe content

https://www.theregister.com/2024/07/30/apple_google_tpu_ai/…

• Meta shuts down celebrity lookalike AI chatbots, focuses on AI Studio tool for creating personal chatbots
• Meta spokesperson Liz Sweeney confirms discontinuation of AI characters embodied by celebrities
• Initial list of chatbots included personas for celebrities like Charli D’Amelio, Dwyane Wade, and Paris Hilton
• Meta paid celebrities millions for their likenesses, but the project did not follow through as planned
• CEO Mark Zuckerberg sees promise in AI chatbots and has integrated Meta AI assistant into various platforms

https://www.theverge.com/2024/7/30/24209918/meta-celebrity-lookalike-ai-chatbots-moves-on…

• OpenAI debuts advanced voice AI feature for ChatGPT platform, utilizing GPT-4o model for hyper-realistic audio responses.
• Feature allows real-time, interruption-capable conversations, addressing challenges in achieving lifelike AI interactions.
• Alpha version released to select ChatGPT Plus subscribers, with plans for wider rollout to premium users in fall.
• Controversy arose over AI voice demo resembling Scarlett Johansson’s voice, leading to legal concerns and removal of demo.
• OpenAI limits system to four preset voices by paid voice actors to prevent misuse and deceptive deepfakes, with filters to block copyrighted content requests.

https://www.pymnts.com/artificial-intelligence-2/2024/openai-debuts-advanced-voice-ai-for-subscribers/…

• The new Siri in iOS 18.1 beta 1 introduces Apple Intelligence, enhancing context understanding and improving user experience.
• Siri in iOS 18.1 features a new visual design with a glowing light effect and better haptic feedback.
• The new Siri allows for typed requests on iPhone and iPad by double-tapping the Home bar, expanding interaction options.
• Siri in iOS 18.1 beta 1 is knowledgeable about Apple products and software features, offering how-to assistance based on Apple support documentation.
• Future updates will bring more enhancements to Siri with Apple Intelligence, promising ongoing improvements beyond the current beta version.

https://9to5mac.com/2024/07/30/this-is-the-new-siri-with-apple-intelligence-in-ios-181-beta-1/…

• OpenAI has released a preliminary version of its voice assistant to paid subscribers, excluding the Scarlett Johansson-like voice that caused controversy.
• The release was delayed due to safety concerns and the need to reach “launch standard,” with plans to expand the feature to all GPT-Plus users in the fall.
• The company faced criticism for featuring a voice similar to Johansson’s, named “Sky,” during a demo in May, which led to privacy features being added to prevent voice impersonation and copyright issues.
• OpenAI has not disclosed the identity of the voice behind “Sky” to protect privacy, and plans to introduce more voice options in the future.
• OpenAI CEO Sam Altman, with a net worth of $1 billion from investments in companies like Reddit and Stripe, founded OpenAI after previous ventures in social mapping and startup acceleration.

https://www.forbes.com/sites/kirkogunrinde/2024/07/30/openai-begins-releasing-chatgpt-voice-assistant-without-scarlett-johansson-like-voice/…

• AI Search War Analysis: Tech giants racing to dominate generative AI-powered search engines
• Tech companies forming revenue-sharing deals with media outlets for AI-generated content
• AI companies treating human content as information mines, leading to less traffic for media websites
• Publishers gaining leverage over AI companies through content control and legal actions
• Uncertainty over the future of AI content usage and publisher partnerships in the evolving search war landscape

https://www.theatlantic.com/technology/archive/2024/07/perplexity-ai-search-media-partners/679294/…

• Advanced spoken conversations online are coming to ChatGPT
• OpenAI introduced an updated voice mode for ChatGPT-4o, featuring lifelike responses and emotional recognition
• The advanced voice mode will be rolled out to paid users, starting with a small group of “Plus” mode subscribers
• Concerns arise regarding the tool’s ability to understand diverse speech patterns and user trust in a human-sounding AI assistant
• OpenAI implemented safety measures for the voice mode, including restrictions on voice options and content generation

https://www.cnn.com/2024/07/30/tech/openai-chatgpt-advanced-voice-mode-rollout/index.html…

• OpenAI introduces Her-like voice mode for ChatGPT Plus subscribers
• Voice mode demonstrated at GPT-4o launch event, delayed for safety improvements
• Advanced mode allows interruptions and adapts responses accordingly
• Model tested by over 100 external red teamers to enhance safety measures
• New voice mode restricted to four preset voices to prevent impersonation

https://www.theverge.com/2024/7/30/24209650/openai-chatgpt-advanced-voice-mode…

• Meta’s Mark Zuckerberg discusses AI with Nvidia’s Jen-Hsun Huang, covering topics like generative systems and smart glasses
• Zuckerberg showcases Meta’s Segment Anything Model 2, demonstrating its object tracking capabilities with a video of cows
• The CEOs’ conversation lacked focus and audience engagement, with attempts at humor falling flat
• Huang jokes about Zuckerberg’s cows and their cooking skills, but fails to generate much reaction
• Despite efforts to engage the audience, the discussion veered away from insightful AI trends and computer graphics, leaving some disappointed

https://www.pcgamer.com/software/ai/metas-mark-zuckerberg-chews-ai-cud-with-nvidias-jen-hsun-huang-talking-about-delicious-cows-slicing-tomatoes-and-old-chinese-guys-drinking-whiskey/…

• Mark Zuckerberg and Jensen Huang envision a future where everyone has an AI assistant, discussing this at SIGGRAPH conference
• Meta’s new platform, AI Studio, allows users to create AI characters to act as assistants for various tasks
• Meta aims to empower users, from creators to small businesses, to create their own AI agents for different purposes
• Jensen Huang highlights the impact of AI on jobs, stating that every job, including his own, will be influenced by AI assistants
• Both tech leaders emphasize the importance of open-source AI, with Zuckerberg expressing a preference for open-source to continue building technology

https://qz.com/mark-zuckerberg-jensen-huang-meta-nvidia-ai-assistants-1851608544…

• Meta AI data centers house around 600,000 Nvidia H100 GPUs: SIGGRAPH 2024
• Nvidia CEO Jensen Huang and Meta CEO discussed importance of open-source technology
• Meta’s Llama 3.1, the largest language model, is open source and aimed at enterprises
• Generative AI will drive the next wave of software creation, according to Jensen Huang
• Nvidia’s Blackwell rollout to Google Cloud expected in early 2025

https://seekingalpha.com/news/4130108-meta-ai-data-centers-hold-about-600000-nvidia-h100-gpus-siggraph-2024…

• Meta introduces AI Studio on Instagram, allowing users in the US to create AI versions of themselves for communication with followers.
• AI profiles can engage with users in chat threads, respond to comments, and be customized based on content preferences and interaction limits.
• The AI characters created can be used across Meta’s apps and shared with others, similar to existing chatbot services like Character.AI and Replika.
• Meta’s approach includes clear labeling of AI profiles and guidelines for creators to specify topics to avoid for their AI versions.
• The company is cautious about potential issues with AI-generated content and emphasizes the creator’s responsibility in managing their AI’s interactions.

https://www.theverge.com/24209196/instagram-ai-characters-meta-ai-studio-release…

• Apple emphasizes a ‘responsible’ training approach for Apple Intelligence models, using a mix of publicly available and licensed data, excluding private user data.
• The technical paper reveals the development of Apple Foundation Models (AFM) for generative AI features in iOS, macOS, and iPadOS.
• Training data for AFM models includes licensed data from publishers, open-source code, and math questions from various sources.
• Apple filters code repositories for minimal usage restrictions and sources additional data for model fine-tuning.
• The paper aims to position Apple as an ethical player in AI development while navigating legal complexities and competitive pressures.

https://techcrunch.com/2024/07/29/apple-says-it-took-a-responsible-approach-to-training-its-apple-intelligence-models/…

• Meta’s latest video vision AI, Segment Anything 2, showcased by CEO Mark Zuckerberg at SIGGRAPH, advances segmentation technology in images and videos.
• The model, Segment Anything 2, identifies and outlines objects in videos efficiently, a significant improvement in computational efficiency in the industry.
• Meta’s AI model, Segment Anything 2, is open and free to use, supported by a large annotated database of videos for training purposes.
• Meta’s commitment to openness in AI, exemplified by models like LLaMa and Segment Anything, aims to enhance the ecosystem rather than solely altruistic intentions.
• The model’s release marks a significant step forward in AI performance, providing a valuable resource for researchers and developers in the field.

https://techcrunch.com/2024/07/29/zuckerberg-touts-metas-latest-video-vision-ai-with-nvidia-ceo-jensen-huang/…

• Nvidia CEO Jensen Huang introduces the concept of physical AI at SIGGRAPH 2024, emphasizing the need for three computers to create, send commands, and handle AI for robots.
• Nvidia unveils new offerings like NVIDIA NIM and OSMO NIM microservices, MimicGen NIM, and Robocasa NIM to accelerate humanoid robotics development.
• Huang showcases an AI-powered customer service agent with human-like interactions, sparking ethical discussions on its realism and potential benefits for individuals.
• Huang envisions a future where personal AIs mimic their human tutors in various industries, with every employee having their own AI assistant.
• Huang highlights the role of AI in enabling innovative designs and tools at Nvidia, emphasizing the augmentation of every job within the company.

https://seekingalpha.com/news/4129838-nvidia-begins-ushering-in-next-wave-of-ai-physical-ai…

• Our new AI model, Segment Anything Model 2 (SAM 2), can segment objects in video in real-time, addressing challenges like fast movement and appearance changes.
• The original Segment Anything Model enabled diverse applications in science, medicine, and industries like marine science, disaster relief, and medical imaging.
• SAM 2’s capabilities extend to easier video editing, mixed reality experiences, and aiding in computer vision system training, including for autonomous vehicles.
• The model allows for tracking objects in videos, real-time interaction, and faster annotation of visual data.
• Meta is sharing its research on SAM 2 to encourage exploration of new capabilities and use cases within the AI community.

https://about.fb.com/news/2024/07/our-new-ai-model-can-segment-video/…

• Machine unlearning is a developing field of research focused on removing specific data from trained AI models
• Deep learning models in AI have the ability to memorize training information and repeat it, posing potential problems
• Machine unlearning aims to produce a model close to a completely retrained one, minus problematic data
• Various techniques for machine unlearning exist, such as exact unlearning, strong unlearning, and weak unlearning
• Machine unlearning may become increasingly important due to privacy concerns and regulatory requirements in the AI landscape

https://www.xda-developers.com/what-is-machine-unlearning/…

• Unlearning techniques present challenges for advanced AI models such as GPT-4o and Llama 3.1 405B, as reported by a collaborative study among institutions like the University of Washington, Princeton, the University of Chicago, USC, and Google.
• Generative AI models function based on learned patterns from vast amounts of data, lacking intent and relying on statistical analysis for responses.
• Unauthorized data scraping by AI model developers has led to copyright disputes, prompting the rise of unlearning techniques to correct models.
• Unlearning strategies aim to enhance data privacy but can impact model performance, necessitating further research for improvement.
• Balancing data privacy and model performance is a key dilemma in generative AI models, with unlearning techniques offering enhanced privacy but also posing challenges in implementation and ethical considerations.

https://elblog.pl/2024/07/29/challenges-in-training-generative-ai-models/…

• Unlearning techniques aim to make AI models forget specific undesirable data, impacting their performance
• Current unlearning methods can degrade models like GPT-40 and Llama 3.1 405B, making them less capable
• Generative AI models learn patterns from vast datasets but lack real intelligence
• Unlearning is crucial due to copyright disputes and privacy concerns in AI model training
• Challenges exist in developing effective unlearning methods, highlighting the need for further research

https://techcrunch.com/2024/07/29/making-ai-models-forget-undesirable-data-hurts-their-performance/…

• Apple introduces Apple Intelligence AI platform with iOS 18.1 developer beta
• Developer betas for iOS, iPadOS, and MacOS Sequoia include Apple Intelligence features
• Apple’s developer program allows early access to upcoming software versions for testing
• Apple plans to launch Apple Intelligence in October, potentially delaying its debut with iPhone 16
• Apple Intelligence is expected to boost iPhone sales, but successful rollout is crucial

https://finance.yahoo.com/news/apple-gives-developers-first-taste-of-apple-intelligence-with-ios-181-developer-beta-171201605.html…

• Apple’s AI features, including Siri enhancements and AI-powered image generation, are expected to debut with iOS 18.1 in October, not with the initial iOS 18 release alongside the iPhone 16 launch in September.
• The unveiling of Apple Intelligence took place at the Worldwide Developers Conference in June, showcasing advancements in Siri, AI-generated images, and integration with ChatGPT.
• The AI features may not be immediately available on new iPhone 16 devices but could be later updated using Presto, Apple’s in-the-box updating system.
• Despite the usual practice of not releasing betas for follow-up software updates before the initial release, Bloomberg reports that some AI features will be included in the iOS 18.1 and iPad 18.1 developer betas.
• Developers will have the opportunity to test and identify bugs in Apple’s AI features early, with the potential for expansion to other Apple platforms like Mac, Apple Watch, and Apple TV.

https://www.theverge.com/2024/7/29/24208656/apple-intelligence-ai-ios-18-1-iphone-16-launch…

• Apple to release artificial intelligence features in October, after iOS 18 update
• Features will work on iPhone 15 Pro models and newer
• ChatGPT integration and Siri improvements expected later in the year
• Developers to get early access in iOS 18.1 beta version
• AI features can proofread, rewrite, create custom emojis, and transcribe phone calls

https://www.cnbc.com/2024/07/29/apple-intelligence-ai-to-release-in-october-report-.html…

• Apple’s AI features delayed until October due to bug fixes and third-party developer support
• Features such as smarter Siri and Genmoji were missing from iOS 18 beta launch in July
• AI features expected to be introduced in iOS 18.1 and iPadOS 18.1 developer betas
• Apple to launch iOS 18, iPadOS 18, and macOS Sequoia in September without AI features
• Apple joins other tech companies in signing Biden administration’s AI code of practice

https://www.yahoo.com/tech/apples-ai-features-report-claims-075623326.html…

• Apple Intelligence release delayed, potential delay to October with iOS 18.1
• Samsung’s Galaxy AI reaches 200 million devices, rapid rollout and compatibility with older devices
• Apple Intelligence features won’t all be available initially, Siri’s AI capabilities may be delayed
• Apple Intelligence requires A17 Pro devices, limited compatibility to iPhone 15 Pro and Pro Max
• Samsung acquires Oxford Semantic Technologies to enhance on-device AI features

https://www.sammobile.com/news/apple-intelligence-release-delayed-while-galaxy-ai-hits-200-million-devices/…

• Apple’s new AI features, known as Apple Intelligence, are delayed after the release of iOS 18.
• The delay is due to concerns about the stability of Apple Intelligence features.
• Apple plans to launch Apple Intelligence in October, after the initial release of iOS 18 and iPadOS 18 in September.
• The company will make Apple Intelligence available to developers for testing on beta versions of iOS and iPadOS 18.1.
• Apple Intelligence uses generative AI to create text, images, and other content on command, and will be available on specific Apple devices.

https://www.foxbusiness.com/technology/apples-new-ai-features-delayed-after-ios-18-release-report…

• AMD launches Ryzen AI 300 Zen5 “Strix Point” mobile series, first reviews now available
• AMD introduces three new SKUs: Ryzen AI 9 HX 370, Ryzen AI 9 365, and Ryzen AI 9 HX 375 with faster NPU units
• New architecture includes more CPU and GPU cores, along with a faster XDNA unit
• Laptop makers can configure a single SKU for power ranges from 15 to 54W, causing potential confusion for customers
• Initial tests show RDNA 3.5 graphics performance, battery life comparable to Qualcomm Snapdragon X Series, challenges with XDNA NPU, limited availability of high-performance Strix Point APUs, and pricing starting at $1399

https://videocardz.com/newz/amd-launches-ryzen-ai-300-zen5-strix-point-mobile-series-first-reviews-now-available…

• AMD Ryzen AI 300 (Strix Point) family introduces next-generation AI capability with a powerful NPU capable of 50 TOPS, surpassing Qualcomm and challenging Intel’s upcoming processors.
• The AMD Ryzen AI 9 HX 370 processor in the Asus Zenbook S 16 is the most advanced silicon from AMD, featuring a 4nm process node and architectural upgrades for enhanced performance.
• The chip architecture includes Zen 5 with improved core capabilities, increased core count, and thread handling, offering a 16% boost in instructions per cycle compared to the previous generation.
• The integrated Radeon 890M graphics in the Ryzen AI 9 HX 370 processor provides enhanced gaming performance and efficiency, supporting 1080p gaming without additional discrete graphics hardware.
• The NPU in the 300 series chips, utilizing XDNA2 architecture, delivers significant power efficiency and performance gains for AI tasks, enabling faster and smarter AI performance across various applications.

https://www.pcmag.com/news/strix-point-first-tests-amd-ryzen-ai-300-laptop-chip-flexes-real-cpu-npu…

• Synchron, a leading-edge brain implant company, integrates OpenAI’s ChatGPT into its software for easier device control for people with paralysis
• ChatGPT assists users like Mark, diagnosed with ALS, in faster communication by providing menu options based on context
• Synchron’s CEO, Tom Oxley, explores AI integration possibilities, including ChatGPT-4o’s multi-input capabilities
• Synchron’s implant, called stentrode, connects the brain to devices by interpreting thoughts for actions wirelessly
• Synchron aims for FDA approval for its BCI, offering hope for those with paralysis like Mark and encouraging participation in finding solutions

https://www.cnet.com/tech/computing/how-this-brain-implant-is-using-chatgpt/…

• OpenAI is testing SearchGPT, a new AI search experience to compete directly with Google
• CrowdStrike offers partners $10 Uber Eats gift cards for global IT outage caused by botched update
• Wiz turns down $23 billion acquisition offer from Alphabet to remain independent
• Elon Musk announces Tesla will sell Optimus humanoid robot in 2026
• Apple Maps now accessible from browser in new public beta

https://techcrunch.com/2024/07/27/openai-comes-for-google-with-searchgpt/…

• Elon Musk’s X has introduced a default setting allowing user posts to train the AI chatbot Grok, with an option to opt out.
• Users can control whether their public posts are used to train Grok, and making accounts private can prevent content from being fed to the AI.
• Grok, a chatbot similar to OpenAI’s ChatGPT, is part of Musk’s AI startup xAI, acquired after Musk paid $44 billion for Twitter in 2022.
• Musk has shown optimism for AI in his business ventures, including SpaceX, but also signed an open letter in 2023 warning of AI risks.
• The letter, supported by prominent scientists including Apple’s Steve Wozniak, called for a pause on the development of AI systems more powerful than GPT-4.

https://deadline.com/2024/07/elon-musk-x-ai-chatbot-grok-1236024077/…

• Elon Musk’s X platform is under pressure from regulators over data harvesting for Grok AI
• Default setting on the app allows user posts to be used for training AI chatbot
• UK and Irish data watchdogs have contacted X over the issue of gaining user consent without their knowledge
• Regulators express concern over the violation of UK and EU GDPR rules regarding default consent methods
• Large language models like Grok are fed data scraped from the internet to build AI, facing opposition from various sectors

https://www.theguardian.com/technology/article/2024/jul/26/elon-musks-x-under-pressure-from-regulators-over-data-harvesting-for-grok-ai…

• OpenAI made a mistake with its new SearchGPT tool, providing incorrect information in a demo
• SearchGPT is an AI tool meant to revolutionize internet search by providing instant, personalized answers
• AI search tools like SearchGPT often exhibit errors such as inventing information or misattributing details
• OpenAI’s SearchGPT aims to offer instantaneous summaries and tailored lists for users
• The demo of SearchGPT showed inaccuracies in providing information about music festivals in Boone, North Carolina

https://www.theatlantic.com/newsletters/archive/2024/07/openais-search-tool-has-already-made-a-mistake/679264/…

• Users on X, formerly known as Twitter, can opt out of allowing their activity to be shared with AI assistant Grok through the “More” tab
• The setting to share data with Grok appeared Thursday night or early Friday morning, with no option to disable it on X’s mobile app
• Users who opt in will allow their public posts to train Grok, which is limited to premium subscribers
• Elon Musk’s AI startup xAI prioritizes artificial intelligence, with Grok being open-source and available to Premium and Premium+ subscribers
• X does not sell user data produced by Grok interactions; Musk’s net worth is estimated at $234.3 billion, making him the wealthiest person

https://www.forbes.com/sites/antoniopequenoiv/2024/07/26/x-quietly-begins-training-ai-assistant-grok-with-user-posts-how-to-disable-the-feature/…

• Opting out of X’s AI training on user data is crucial to protect your privacy and data
• X, formerly Twitter, has introduced a default setting that allows its AI platform to learn from user data unless changed
• The platform has updated its terms of service to allow for the collection of biometric data and employment history for AI training
• Concerns have been raised about companies training AI on user data without explicit permission
• To opt out of X’s Grok training, users must access the desktop version, go to Settings, select Privacy and safety, and choose Grok under Data sharing and Personalization

https://www.cnet.com/tech/services-and-software/x-is-using-your-tweets-to-train-its-ai-heres-how-to-disable-that/…

• Stop Grok AI data: Twitter users automatically opted in to share data with Grok AI tool
• Grok AI tool can access and use posts without permission, even for non-Grok users
• Users can easily disable data sharing with Grok AI through web settings
• Concerns raised about reliability of Grok AI tool in generating false information
• Grok AI’s impact on spreading misinformation remains uncertain

https://www.fastcompany.com/91163676/x-gave-grok-ai-automatic-permission-to-train-using-your-data-heres-how-to-stop-it…

• Elon Musk’s privacy concerns arise as X, formerly Twitter, defaults user data into Grok AI training pool
• Grok, a conversational AI developed by Elon Musk-owned X, is intended as a rival to OpenAI’s ChatGPT
• European privacy watchdog, the Irish Data Protection Commission, expresses surprise at X’s move
• Unclear if X is using all user data or only interactions with Grok for training purposes
• Similar plan by Meta on repurposing data for AI training paused in Europe due to GDPR complaints

https://techcrunch.com/2024/07/26/privacy-watchdog-says-its-surprised-by-elon-musk-opting-user-data-into-grok-ai-training/…

• OpenAI’s SearchGPT aims to challenge Google’s dominance in the search engine space
• SearchGPT combines AI models with web data for quicker, up-to-date answers and relevant sources
• OpenAI plans to gather feedback on SearchGPT from users and publishers
• Google recently updated its search engine to include AI-generated summaries at the top of results
• Microsoft, a partner of OpenAI, is also testing AI summaries on its Bing search engine

https://www.fox10phoenix.com/news/openai-searchgpt-google-search-engine…

• Elon Musk’s X platform now automatically trains Grok on user data, but users can opt out through the website settings.
• Concerns have been raised about companies training generative AI tools on user data without explicit permission.
• Grok was recently announced as the most powerful AI training cluster in the world by Musk.
• Users discovered the default setting allowing data training on X through a post on the platform.
• Companies like Meta and OpenAI also use publicly accessible user data to train their AI models for more human-like interactions.

https://www.zdnet.com/article/elon-musks-x-now-trains-its-grok-ai-on-your-data-by-default-heres-how-to-opt-out/…

• To disable Twitter data training for Grok AI, users can switch off the setting that allows X to train its AI on users’ posts
• X, formerly known as Twitter, activated a new setting by default that allows it to use user data for training purposes
• Users can delete their conversation history with the AI after disabling the setting
• X owner Elon Musk mentioned that xAI had started training its AI using a powerful training cluster
• Meta also planned to use public content on Facebook and Instagram to train its AI but paused the plans due to regulatory pressure

https://techcrunch.com/2024/07/26/heres-how-to-disable-x-twitter-from-using-your-data-to-train-its-grok-ai/…