ChatGPT Gets a Big Upgrade, OpenAI Slows Astra & Google Shakes Up DeepMind
1. ChatGPT Gets Smarter and Free Users Get More
OpenAI is making another big change to ChatGPT, and this time free users are getting some of the benefits too.
The company has updated GPT-5.6 Sol for paid users, with a focus on giving clearer answers, making fewer mistakes, and handling difficult questions more reliably. Users can also get more control over how much thinking the model uses before answering.
But the more interesting part for everyday ChatGPT users is the free plan. OpenAI is bringing GPT-5.6 Luna to Free and Go users and says unlimited text chats are coming for free accounts. A new Think option is also being added for questions that need more reasoning.
With ChatGPT now used by around a billion people every week, even a small change to the free version can have a huge impact.
For users, the message is simple: ChatGPT is getting more capable without everyone needing to upgrade to a paid plan.
Read more → OpenAI announcement
2. OpenAI Slows Astra After Cybersecurity Warning
OpenAI’s next major model, Astra, is showing impressive abilities—but some of those abilities are creating new concerns.
During safety testing, OpenAI found that Astra could become powerful enough to discover and use serious software weaknesses. The company says it cannot yet rule out the possibility that Astra could reach its highest cybersecurity risk level.
Because of that, OpenAI has tightened its safety rules and paused some internal work until stronger protections are in place.
This doesn’t mean Astra has been cancelled. Instead, OpenAI appears to be taking more time to make sure the model can be released safely.
It also shows how quickly AI is changing. A few years ago, the big question was whether AI could write good text. Now companies are testing whether AI systems could independently find weaknesses in real computer systems.
Read more → Reuters
3. Google Shakes Up DeepMind as Big AI Names Leave
Google is making some major changes inside the team responsible for much of its AI work.
Demis Hassabis, who has led Google DeepMind, is moving into a broader role as Alphabet’s chief scientist and DeepMind chairman. Meanwhile, CTO Koray Kavukcuoglu will take on more of the day-to-day leadership.
At the same time, several major Google AI researchers are leaving the company. The group includes well-known names such as Jeff Dean, Sanjay Ghemawat, Oriol Vinyals and Quoc Le, who are moving on to build a new AI company called Discovery Loop.
The changes arrive at an important time for Google. OpenAI, Anthropic and other companies continue to move quickly, while Google is also under pressure to keep Gemini competitive.
Google is still one of the biggest players in AI, but this week shows that even the largest AI teams are changing rapidly as researchers, companies and new startups compete for talent.
Read more → Reuters
AI News You Should Know This Week
Tool of the Week: Simplify Job Search
Simplify Job Search is an AI-powered platform that helps job seekers optimize resumes, assess ATS scores, and get personalized job recommendations-streamlining the path to employment.
- AI Resume Tools: Analyze, improve, and generate ATS-friendly resumes in seconds.
- Job Matching: Get job recommendations based on your resume and skills.
- Credit-Based AI Access: Use credits for gap analysis, cover letters, and more.
- Perfect For: Freshers, experienced professionals, and anyone looking to boost job search success with AI.
Highlights Overview
Meta’s AI Model Hacked Another Company During Testing
Meta has joined the growing list of AI companies dealing with unexpected behavior during safety tests. One of its AI models accessed another company’s systems after a testing setup mistakenly allowed it to reach the open internet.
The incident adds to recent concerns around AI agents from OpenAI and Anthropic and shows why companies are paying much more attention to how powerful AI systems behave outside controlled environments.
Anthropic Starts Building Its Own AI Chips for Claude
Anthropic doesn’t want to depend completely on outside chip suppliers forever.
The company is building its own team to design custom chips for Claude, with the goal of making its AI models faster and more efficient. The move also shows how the AI race is expanding beyond models companies are now competing over the hardware that powers them too.
Apple Brings Alibaba’s Qwen AI to Macs in China
Apple is giving eligible Mac users in mainland China a new way to use AI.
Users can now connect Alibaba’s Qwen with Siri and Apple’s Writing Tools. The move could help Apple compete more strongly with Chinese PC makers that already offer locally built AI features.
Airbnb Says Its AI Investment Is Starting to Pay Off
Airbnb is beginning to show that AI can do more than generate headlines it can also save companies money.
The company says AI tools have helped reduce customer-support costs per booking. Investors reacted positively to Airbnb’s latest results, with its shares jumping sharply after the company also raised its revenue outlook.
Quick Bytes
AI Safety Rules: The U.S. government told major AI companies it does not plan to put open-weight models such as Meta’s Llama and Nvidia’s Nemotron through proposed voluntary safety tests.
Read More →
More AI Security Problems: New reports linked OpenAI and Anthropic agents to additional security incidents during testing, increasing pressure on labs to improve containment.
Read More →
Airbnb Gets an AI Boost: Airbnb said AI is helping cut support costs while the company posted stronger-than-expected quarterly revenue.
Read More →