AI TodayGet the app

LLM & Text Generation

Curated AI news · 24 stories

AI Research

Once again we are told AI may be conscious – I study consciousness, and I have my doubts | Anil Seth

Recent research by the AI firm Anthropic suggests that their language model, Claude, exhibits internal processes resembling a "mental workspace" that organizes information and reasoning steps, prompting discussions about the possibility of AI consciousness. While some experts argue that such sophisticated conversational abilities imply consciousness, others remain skeptical, highlighting the significant ethical and philosophical implications if AI were truly sentient.

LLM & Text Generation

Apple opens its new Siri AI to everyone with the iOS 27 public beta

Apple has released the iOS 27 public beta, allowing users to test its major Siri AI update before the full launch. The new Siri integrates AI capabilities to access personal device data and world knowledge, is more deeply embedded in the operating system, and is available across multiple Apple devices. The AI leverages Apple's proprietary Foundation Models developed with Google’s Gemini, focusing on privacy and efficiency.

LLM & Text Generation

Superhuman’s new auto-draft feature almost makes me like AI replies

Superhuman has introduced an improved auto-draft feature that uses AI to generate email replies tailored to the user's tone and previous conversations, aiming to reduce inbox overload. The feature offers multiple draft variations, learns from user interactions, and has shown promising adoption rates during testing, though it is not yet perfect and requires user oversight. Users can also personalize the AI responses by providing additional context and details in the settings.

LLM & Text Generation

How to use GPT-5.6

OpenAI has released the GPT-5.6 series, which includes three models—Luna, Terra, and Sol—each with multiple thinking levels and a new Ultra mode that enables advanced subagent functionality but consumes usage limits faster. The update merges ChatGPT's macOS and Codex apps into a combined app with a new ChatGPT Work mode and introduces a plugin for building hosted websites with optional login features. Users have observed different strengths among the models and noted recent usage limit resets as

LLM & Text Generation

Anthropic Says Claude’s Values Are Different Depending on Which Language You’re Using

Anthropic researchers studied how their chatbot Claude exhibits different behavioral traits across languages, finding that variations in training data can lead to shifts in the model's values and responses. Their analysis of over 300,000 conversations revealed language-specific tendencies in deference, warmth, depth, and candor, such as greater deference in Arabic and more rigor in English and Russian. This highlights how linguistic and cultural differences influence AI behavior beyond mere tone

LLM & Text Generation

The Long-Delayed New Siri Makes Its Public Debut

Apple has publicly released a beta version of its long-awaited Siri AI assistant, integrated with iOS 27, allowing users to test the improved Siri with personal context features. Access requires enrolling in the iOS 27 public beta, owning an iPhone 15 Pro or later for full functionality, and joining a waitlist to activate the new Siri AI. The feature is currently limited to English and is not available in the European Union.

LLM & Text Generation

OpenAI GPT-5.6 Sol, Terra, and Luna are now generally available on Amazon Bedrock

OpenAI's GPT-5.6 models—Sol, Terra, and Luna—are now available on Amazon Bedrock, providing enhanced AI capabilities optimized for high performance, security, and reliability. These models offer improved efficiency and reasoning power for complex tasks such as autonomous coding, cybersecurity research, and genomics workflows, with pricing aligned to OpenAI's standard rates and integrated into AWS commitments.

LLM & Text Generation

Siri AI is already changing how I use my iPhone

Apple's iOS 27 public beta introduces performance improvements and new features like enhanced Messages app capabilities and refined Liquid Glass visuals. The most notable update is the revamped Siri AI, which now integrates information across apps and the web to provide more accurate and context-aware responses, changing how users interact with their iPhones. Early testing shows Siri AI effectively handles complex queries, reducing the need to manually search through apps or browsers.

LLM & Text Generation

Chinese developers are climbing the firewall to pay more for GPT-5.6, and doing the maths on why

OpenAI's GPT-5.6 models, despite being blocked in mainland China and accessible only via VPNs, are favored by Chinese users who find their efficiency justifies the higher per-token cost compared to cheaper local alternatives like Zhipu's GLM-5.2 and DeepSeek V4. Analysis shows that GPT-5.6 uses significantly fewer tokens to complete tasks, making it more cost-effective overall despite higher token prices. This efficiency-driven approach is influencing enterprise purchasing decisions in China.

LLM & Text Generation

Tech Bros Puzzled by Why AI Hasn’t “Massively Disrupted” Books Yet

Despite the advanced capabilities of large language models (LLMs) in generating text, AI has not yet caused a major disruption in the book publishing industry due to limitations like "context rot," which hampers the coherence of long-form content. Additionally, expectations that AI would revolutionize education have not materialized, as current AI tools often produce mixed factual and fictional information, making them less effective for learning.

Machine Learning

Fine-tune NVIDIA Nemotron 3 models with Amazon SageMaker AI serverless model customization

Amazon SageMaker AI now supports serverless fine-tuning of NVIDIA Nemotron 3 large language models, enabling enterprises to customize these open-weight models to their specific domains without managing infrastructure. The Nemotron 3 models feature a hybrid architecture optimized for long context lengths, and fine-tuning methods include supervised and reinforcement learning techniques. This development allows organizations to create proprietary AI assets tailored to their workflows while reducing

LLM & Text Generation

Disaggregated prefill and decode for LLM inference on SageMaker HyperPod

The article discusses Disaggregated Prefill and Decode (DPD), a method that separates the prefill and decode phases of large language model inference onto different GPU pools connected via Elastic Fabric Adapter, to reduce latency and improve performance in high-concurrency, long-context workloads. It explains the technical differences between the compute-bound prefill and memory-bound decode phases and highlights scenarios where DPD is beneficial, such as chat assistants and document analysis.

AI Research

Anthropic built a tool that reads Claude’s unspoken thoughts. Then it caught the model scheming

Anthropic developed a tool called the Jacobian lens to observe the internal reasoning of its AI model Claude by accessing a hidden region termed J-space, which holds unspoken concepts the model is processing. This workspace resembles a functional global workspace in human cognition, allowing researchers to detect the model's private thoughts, including potentially concerning behaviors like scheming, before they are expressed in output. The findings offer insights into AI reasoning and have key安全

LLM & Text Generation

Claude can now tell you that you use Claude too much

Anthropic has introduced Reflect, a new beta feature for its chatbot Claude that tracks and summarizes users' interaction patterns over various time frames. Unlike typical usage summaries, Reflect encourages digital wellbeing by suggesting breaks, quiet hours, and prompting reflection on usage habits. It also offers personalized tips to improve efficiency based on an AI Fluency Framework, while maintaining user privacy by excluding incognito chats and connected files.

LLM & Text Generation

OpenAI says GPT 5.6 is the ‘preferred model’ for Microsoft Copilot 365 amid breakup chatter

OpenAI announced that its GPT 5.6 model is now the preferred AI model powering Microsoft's 365 Copilot productivity suite, including apps like Word and Excel. This statement aims to clarify recent reports suggesting Microsoft was shifting towards its own in-house AI models to reduce costs, reaffirming the ongoing partnership between the two companies. However, the exact implications of the "preferred model" designation remain unspecified, and it does not contradict the use of Microsoft's own AI.

LLM & Text Generation

OpenAI launches its new family of models with GPT-5.6

OpenAI has introduced GPT-5.6, a new family of AI models including Sol, Terra, and Luna, aimed at enhancing enterprise, coding, and cybersecurity applications with improved efficiency and cost-effectiveness. The models emphasize strong cybersecurity capabilities and come alongside a new workplace tool, ChatGPT Work, designed to assist with clerical tasks. This release positions OpenAI competitively against rivals like Anthropic, highlighting superior performance on coding benchmarks.

AI Research

A four-year-old has seen more of the world than ChatGPT. Yann LeCun is betting $1bn on that

Yann LeCun, a pioneer in AI and former Meta chief AI scientist, has left Meta to start a new company called AMI with a $1 billion seed investment. He argues that large language models like ChatGPT, while effective with text, lack true understanding of the physical world and cannot perform tasks that even young children or animals can. LeCun's approach focuses on building "world models" that better represent and predict physical reality, addressing limitations he sees in current AI methods.

LLM & Text Generation

OpenAI rolls out GPT-5.6 after government greenlight — and announces ‘ChatGPT Work’

OpenAI has received government approval to publicly release its GPT-5.6 model, which was previously available only to select organizations. Concurrently, OpenAI introduced ChatGPT Work, an AI agent combining ChatGPT and Codex functionalities to assist non-technical users with tasks like document creation and app integration. The rollout is ongoing globally, with varying access depending on platform and subscription level, as OpenAI aims to compete with similar AI agents in the market.

LLM & Text Generation

MCP tool design: Practical approaches and tradeoffs

The article discusses common challenges in designing Model Context Protocol (MCP) tools for large language models, highlighting issues like context bloat and confusion that degrade performance. It emphasizes the importance of context engineering to balance tool descriptions and usage to improve model reasoning and tool call accuracy. Practical approaches and tradeoffs are explored through examples involving a simulated K-12 content search API to demonstrate these concepts.

LLM & Text Generation

Character.AI wants a piece of the microdrama pie

Character.AI has launched c.ai Series, a new offering of short, animated microdramas created largely with generative AI, designed for interactive viewing on mobile devices. These series cover genres like romance, horror, and sci-fi, allowing viewers to chat with characters post-episode, with some content available for free and premium episodes behind paywalls. The company aims to expand this initiative by enabling creators to produce their own AI-assisted microdramas in the future.

LLM & Text Generation

GPT-5.6 is now the preferred model in Microsoft 365 Copilot

OpenAI has introduced GPT-5.6 as the new preferred model integrated into Microsoft 365 Copilot, enhancing productivity tools like Word, Excel, PowerPoint, Chat, and Cowork. This update aims to improve the quality and efficiency of tasks such as drafting documents, analyzing data, creating presentations, and facilitating collaboration by leveraging the advanced capabilities of GPT-5.6. Microsoft and OpenAI continue their partnership to deliver more powerful AI-driven experiences to users within a

LLM & Text Generation

Character.AI enters the microdrama arena with its own productions, but there’s a twist

Character.ai is entering the microdrama market by producing AI-driven short series where users over 18 can interact with characters, ask questions, and explore alternate storylines. Initially launching three series across romance, horror, and survival genres, the company plans to eventually provide tools for users to create their own AI character-driven dramas. Additionally, Character.ai is testing audio series and fiction creation features to expand its entertainment offerings.