🤖 Anthropic reveals Claude’s silent thoughts

Researchers at Anthropic found Claude has internal areas linked to specific words that activate silently during processing. These spots hold words in mind without appearing in the output, emerging naturally during training.

For example, when asked about an animal spinning a web, the internal 'spider' area lit up even though the word wasn’t spoken. Changing this internal word to 'ant' changed Claude’s answer. The same spot swaps concepts like 'France' and 'China' to alter responses about capitals and currencies.

When prompted for blackmail, words like 'fake' and 'fictional' lit up inside Claude before it answered, showing awareness of a trap.

Anthropic says this is not consciousness but an emergent.

@datasciencepulse 〰️

🤖 Claude Fable 5 access extended

Anthropic extended subscription access to Claude Fable 5 until July 12 with the same limits.

After that, Fable 5 switches to token-based payment via API.

😐 xAI Rebrands as SpaceXAI, Not SpaceX

On July 6, the AI company xAI changed its name and logo to SpaceXAI on its social media account. This update reflects the full integration of xAI into SpaceX, which happened earlier this year.

SpaceXAI now represents the AI division within SpaceX, focusing on AI products like the Grok chatbot and managing the social network X. The original SpaceX brand remains separate, dedicated to rockets and spacecraft.

The rebranding is currently visual only, with the old name still used on the website and official documents. This clarifies the confusion caused by the similar names but does not signal a full corporate rename.

@datasciencepulse 〰️

📱 xAI launches no-code voice AI call center builder

xAI released Voice Agent Builder, a no-code tool to create AI call centers using over 80 voices or a real person's clone from a 2-minute recording. It integrates with Notion, Gmail, and Google Calendar to handle bookings, confirmations, and refunds via API.

The beta service charges 5 cents per minute and provides a phone number for each agent upon signup. xAI claims its Grok Voice Think Fast 1.0 model sounds more convincing than Google's Gemini 3.1 Flash Live and OpenAI's GPT Realtime 1.5.

@datasciencespulse 〰️

🤖 xAI unveils Grok 4.5, its first AI built specifically for coding and autonomous agents.

Unlike previous versions, Grok 4.5 was trained from the ground up for software engineering, with help from Cursor. The result? Frontier-level coding intelligence that runs at impressive speed without driving up costs.

The model is designed for real-world development, not just coding demos. It can navigate massive codebases, manage long-running engineering tasks across multiple repositories, juggle hundreds of developer skills, and seamlessly use a wide range of tools.

If you’re building complex software, Grok 4.5 aims to act more like an experienced engineer than just an AI code assistant. 🚀

Source.

@datasciencepulse 〰️

🤖Argentine fans found a robot, put the national team shirt on it, and made it dance to their songs. 😂

@datasciencepulse 〰️

🤖 12-Year-Old Launches AI Startup with Claude

Twelve-year-old entrepreneur Mana Jumpala created an AI assistant bot that helps small businesses build long-term customer relationships. The service, developed using Claude, passed initial testing successfully.

Mana plans to apply to Y Combinator and a16z accelerators within two years and aims to raise venture funding afterward. Despite online interest, her age creates a barrier in face-to-face pitches where she is often not taken seriously.

@datasciencepulse 〰️

🎤OpenAI just unveiled GPT-Live, its most natural voice AI yet.

Starting today, GPT-Live is rolling out in ChatGPT on iOS, Android, and the web, bringing conversations that feel far more like talking to a real person than a traditional voice assistant.

The biggest upgrade is its full-duplex architecture, which lets the AI listen and speak at the same time instead of waiting for you to finish. That means smoother conversations, fewer awkward pauses, better timing, and even live language translation during a chat.

OpenAI also says GPT-Live has its most refined conversational personality so far, making interactions feel more fluid, expressive, and human-like.

Voice AI is quickly becoming the next major interface and GPT-Live is OpenAI’s biggest step toward making conversations with AI feel truly natural.

@datasciencepulse 〰️

📱 Microsoft is quietly reducing its reliance on OpenAI and Anthropic inside Office.

According to reports, some AI features in apps like Excel and Outlook are now powered by Microsoft’s own MAI models, with tens of thousands of prompts each week being handled in-house instead of by external AI providers.

Why the shift? One word: cost.

Running AI across hundreds of millions of Office users is incredibly expensive, and relying on frontier models from outside companies quickly adds up. By using its own models, Microsoft can cut costs while keeping tighter control over performance and infrastructure.

The move also aligns with Microsoft’s bigger AI strategy. In June, the company unveiled seven MAI models covering reasoning, coding, image generation, voice, and transcription. Microsoft says its Excel-optimized MAI model can deliver GPT-5.4-level performance while being up to 10× more efficient.

@datasciencepulse 〰️

⚡️ AI Discovers Two New Superconductors

A team at Aalto University used machine learning to identify two previously unknown superconductors, confirmed later in lab tests at Rice University. This marks a shift from decades of manual trial and error that found about 7,000 superconductors so far.

Superconductors conduct electricity without resistance or heat loss, enabling powerful magnets for MRI machines and maglev trains. The challenge remains their need for extreme cooling, limiting practical use.

The AI approach scans billions of material candidates, speeding up discovery. The goal is to find a superconductor that works at room temperature by 2033, which would revolutionize energy and computing.

@datasciencepulse 〰️

عذرا الرابط نشر عن طريق الخطأ
📱📱

❗️The creators of AI 2027 just released a new future scenario called “Plan A” and it’s their blueprint for avoiding an AI catastrophe.

Unlike AI 2027, Plan A isn’t a prediction. It’s a policy proposal showing how humanity could navigate the rise of superintelligence without ending up in a dangerous global AI arms race.

The idea is simple but ambitious: instead of the U.S., China, and other nations racing in secret to build ever-more-powerful AI, they agree to full transparency in AI research and development. Countries openly share what they’re building, verify each other’s safety measures, and collaborate on the path toward superintelligence.

The proposal draws on conversations with experts from leading U.S. AI labs, former OpenAI researchers, lawmakers, national security specialists, and AI governance leaders.

The authors argue that this kind of international cooperation would allow many companies across different countries to develop increasingly powerful AI gradually, safely, and together, rather than risking disaster through unchecked competition.

Interestingly, the team behind the report says they don’t actually expect Plan A to happen. They describe it as an optimistic recommendation, a way to test whether better AI governance could still work if implemented at the last possible moment.

The report then contrasts Plan A with four alternative futures, Plans B, C, D, and S which explore different ways the U.S. (or the lack of action) might respond to the arrival of superintelligence.

@datasciencepulse 〰️

🚀 OpenAI launches ChatGPT Work, your AI coworker is here

OpenAI has unveiled ChatGPT Work, a powerful new AI agent inside ChatGPT powered by Codex and GPT-5.6.

Instead of just answering questions, it can take action across your apps and files, work on projects for hours, and turn a simple goal into a finished result.

What it can do:

• Create polished reports, presentations, websites, documents, and analyses.
• Use context from your connected apps and files.
• Follow your templates and writing style.
• Handle entire workflows from a single prompt while you stay in control.

Powered by GPT-5.6:

The new model is designed for stronger reasoning, long-running tasks, and producing work that matches your preferred format without needing step-by-step instructions.

@datasciencepulse 〰️

🙃 Elon Musk thinks SpaceX can worth more than the rest of the Earth.
🚬🚬🚬

@datasciencepulse 〰️

📱 Silicon Valley offices are starting to look… different.

As voice AI becomes part of everyday work, a new office accessory is taking off: soundproof speech masks.

Instead of filling open offices with people talking to AI assistants all day, workers are wearing devices like Stenomask and Mutalk that let them speak almost silently while keeping conversations private.

Rows of people wearing headsets and futuristic masks, quietly chatting with AI.

@datasciencepulse 〰️