AI News

Follow the latest AI and automation news

$Google launches Gemini 3.8 Live to take on OpenAI's GPT-Live-1 at a fraction of the cost$

15 Sept 20264 Rabiʻ II 1448 AH

Google launches Gemini 3.8 Live to take on OpenAI's GPT-Live-1 at a fraction of the cost

Google DeepMind has launched Gemini 3.8 Live and 3.8 Live Extended Thinking, two new audio models for developers that lead the speech-to-speech analysis leaderboard. Google's pricing at $1.38 per hour for voice conversation significantly undercuts OpenAI's GPT-Live-1, which still offers a more natural sound due to its full duplex technology. This launch marks a significant step for Google in the voice AI space, as decision-makers in the market seek cost-effective solutions while maintaining performance quality. This could influence how companies adopt this technology in the future.

Governance

15 Sept 20264 Rabiʻ II 1448 AH

AI labs have a data trust problem that their policies haven't solved

OpenAI and Anthropic have assured corporate clients that their data will not be used for training. However, when Anthropic announced it would store usage logs from its Fable model for 30 days, companies like Palantir, Nvidia, and Booz Allen Hamilton withdrew from using it for sensitive tasks. This incident highlights a significant data trust issue within AI companies, as existing policies have not adequately addressed these concerns. While organizations strive for transparency, apprehensions remain regarding the handling of collected data and its implications.

Models

15 Sept 20264 Rabiʻ II 1448 AH

Meta expands subscription push with new AI-focused plans

Meta has announced the launch of a new bundle that provides expanded access to its AI tools, featuring premium capabilities across Facebook, Instagram, and WhatsApp. This bundle includes additional features aimed at enhancing user experience, reflecting Meta's commitment to developing and providing AI tools to its users. This move comes at a time when reliance on AI is increasing to improve digital interactions. The new bundle is expected to influence how users engage with social media platforms, making it easier for them to access advanced tools that enhance content creation and audience interaction more effectively.

💡

From Megawatts to Tokens: How NVIDIA Maximizes AI Factory Production

What strategies can enhance energy efficiency in AI factories?

Infrastructure

AI Infra Summit: NVIDIA Vera Rubin and DSX Platform Advancements Showcase Energy Efficiencies of Optimizing Tokens Per Watt for AI Factories

15 Sept 20264 Rabiʻ II 1448 AH

Infrastructure

15 Sept 20264 Rabiʻ II 1448 AH

AI Infra Summit: NVIDIA Vera Rubin and DSX Platform Advancements Showcase Energy Efficiencies of Optimizing Tokens Per Watt for AI Factories

Ian Buck, NVIDIA's VP of high-performance computing, discussed AI factory efficiency at the AI Infra Summit in Santa Clara. The event saw over 8,000 attendees this year, a significant rise from 3,500 last year, where Buck highlighted new collaborations across NVIDIA platforms. Amazon's Annapurna Labs is collaborating with NVIDIA on NVHBM high-bandwidth memory technology, while d-Matrix integrates with NVLink Fusion to combine NVIDIA Vera CPUs with Raptor XPUs. NVIDIA and its partners showcased new results, including a flexible-load program for AI factories in partnership with Silicon Valley Power.

Governance

OpenAI, Anthropic, and Google Collaborate on AI Standards Body

15 Sept 20264 Rabiʻ II 1448 AH

Governance

15 Sept 20264 Rabiʻ II 1448 AH

OpenAI, Anthropic, and Google Collaborate on AI Standards Body

OpenAI announced its collaboration with Anthropic and Google to establish a standards body for testing and auditing AI models. Teams from the three companies have been meeting regularly since July to discuss the creation of a safety and standards body, based on a proposal by Demis Hassabis from Google DeepMind. The idea involves creating a regulatory mechanism that combines government oversight with self-regulation by the industry, funded by AI companies and including independent technical experts. Hassabis suggested that the body initially operate voluntarily, allowing companies to submit their models for testing before release.

Business

Meta Platforms Launches AI Subscription Service 'Meta One'

15 Sept 20264 Rabiʻ II 1448 AH

Business

15 Sept 20264 Rabiʻ II 1448 AH

Meta Platforms Launches AI Subscription Service 'Meta One'

Meta Platforms announced the launch of its AI subscription service 'Meta One' on Tuesday, offering advanced features across its social media applications. The service has already registered 15 million subscribers during its trial phase and provides over 50 features aimed at enhancing content creation and audience interaction. Subscription prices start at $2.99 for individual products, reaching $14.99 for packages aimed at content creators. These packages include professional tools such as account verification and advanced analytics, enhancing users' capabilities in managing their businesses.

Business

Meta’s new One subscriptions put a price on social media and AI

15 Sept 20264 Rabiʻ II 1448 AH

Business

15 Sept 20264 Rabiʻ II 1448 AH

Meta’s new One subscriptions put a price on social media and AI

Meta has announced the launch of new subscription bundles that include usage of its AI assistant, Muse, with global availability starting today. The new bundles offer several tiers for individual users, creators, and businesses while maintaining the core free experience on its apps. Meta has been testing these bundles since earlier this year, now providing diverse options for users, reflecting its commitment to expanding smart services. The company also confirmed plans to expand the bundles to include additional features like editing tools and smart glasses.

Business

After warning AI is too dangerous, Bill Gates bets a billion on its upside

15 Sept 20264 Rabiʻ II 1448 AH

Business

15 Sept 20264 Rabiʻ II 1448 AH

After warning AI is too dangerous, Bill Gates bets a billion on its upside

The Gates Foundation has announced a billion-dollar investment over two years to make AI tools more accessible in health, education, and agriculture. Bill Gates highlighted that over 90% of the training data for early language models came from English sources, revealing a significant gap in technology access. The foundation warns that the market does not guarantee equal opportunities, as speech recognition fails 60% of the time in languages like Yoruba, necessitating greater efforts to ensure fairness in AI utilization.

Infrastructure

Delos Data Secures $100 Million to Develop Data Transfer Chips

15 Sept 20264 Rabiʻ II 1448 AH

Infrastructure

15 Sept 20264 Rabiʻ II 1448 AH

Delos Data Secures $100 Million to Develop Data Transfer Chips

Delos Data, founded by former Intel experts, has raised $100 million in a funding round to develop chips and software that accelerate data transfer in AI data centers. The company aims to tackle the challenges of data transfer as data centers become increasingly complex, relying more on new technologies rather than traditional graphics processing units. Dan Daly, the CTO, emphasizes that the company seeks to provide a fast means of data transfer between AI infrastructure components, reducing energy waste caused by waiting for data.