Your Hub for Tech News and Handy Online Tools.

Artificial Intelligence

OpenAI Fine-Tunes GPT-5 for Friendlier Interactions After Initial User Feedback

OpenAI is updating GPT-5 to be more approachable after user feedback indicated a preference for the previous GPT-4o model.
The update includes subtle changes to make the model feel warmer, such as using phrases like 'Good question' or 'Great start.'
OpenAI assures that these changes do not increase sycophancy compared to the previous GPT-5 personality. Executives acknowledge the rocky launch and are working to improve user experience.

AI & Education

Duolingo CEO Clarifies AI-First Strategy Amid Criticism

Duolingo CEO Luis von Ahn addressed the criticism following his announcement that Duolingo would become an 'AI-first company,' stating his comments lacked context. He clarified that the company's intention was not solely for profit or to lay off human employees.

Von Ahn emphasized that Duolingo has never laid off full-time employees and has no plans to do so. While acknowledging a reduction in the contractor workforce based on fluctuating needs, he maintained a positive outlook on the potential of AI, with Duolingo dedicating time for experimentation with the technology.

Artificial Intelligence & Toys

Doubts Raised Over A.I. Plushies as Child Companions: Replacing Human Interaction?

The New York Times' Amanda Hess expresses reservations about A.I.-powered plush toys like Curio's Grem and Grok, questioning their role as replacements for parental interaction rather than upgrades to traditional toys.

Hess argues that while marketed as screen-time alternatives, these toys may actually encourage children to seek answers and curiosity within technology, potentially isolating them.

Despite initial reservations, Hess allowed her children to play with Grem after removing the voice box, indicating a preference for imaginative play over A.I. interaction.

Artificial Intelligence

OpenAI's Vision Beyond GPT-5: New Apps, Brain-Computer Interfaces, and a Potential AI-Powered Future

OpenAI CEO Sam Altman hosted tech reporters to discuss the company's future, hinting at ventures beyond AI model launches like GPT-5. Fidji Simo will lead the development of consumer apps outside ChatGPT, including a potential AI-powered browser to compete with Chrome. Altman also suggested OpenAI might consider buying Chrome if it were for sale.

OpenAI is reportedly planning to invest in Merge Labs, a brain-computer interface startup challenging Elon Musk's Neuralink. Despite a bumpy GPT-5 rollout, OpenAI's API traffic doubled, and the company is addressing user concerns about the model's tone.

Altman envisions OpenAI expanding into areas like data centers, robotics, and energy, potentially resembling Alphabet. The company is preparing for a possible IPO in the future.

Social Media

TikTok Updates Community Guidelines with Focus on Clarity and Personalization

TikTok is updating its Community Guidelines, set to go live on September 13, 2025, with changes focusing on clarity and personalization.

Key updates include:

  • Rewriting guidelines for simplicity.
  • Emphasizing creator responsibility on TikTok LIVE, especially regarding third-party tools.
  • Requiring disclosure of commercial content and reducing visibility of content directing users to off-platform purchases where TikTok Shop is available.
  • Personalizing search results and comment sections based on user activity.
  • Simplifying language regarding AI content, removing specific mentions of AI endorsements.
  • Distributing details about ineligible content across various sections, instead of a single list.
  • Changing the reason for content moderation from maintaining a "safe, trustworthy, and vibrant" platform to a "safe, fun, and creative" one.

The updates also address compliance with global regulations and aim to provide a more tailored experience for users.

Mobile AI Applications

ChatGPT Mobile App Surpasses $2 Billion in Revenue, Dominates Market

ChatGPT's mobile app has generated $2 billion in consumer spending since its launch in May 2023, significantly outpacing competitors like Claude, Copilot, and Grok.

In 2025 alone, the app has earned $1.35 billion, a 673% increase year-over-year. Monthly revenue averages $193 million, compared to Grok's $3.6 million.

ChatGPT's app has been downloaded approximately 690 million times globally, with an average of 45 million monthly downloads. Spending per download is $2.91, higher than Claude ($2.55), Grok ($0.75), and Copilot ($0.28).

India leads in downloads (13.7%), while the U.S. accounts for the largest share of revenue (38%).

Social Media

Bluesky Revamps Policies & Guidelines: A Safer, More Respectful Social Network?

Bluesky is updating its Community Guidelines and policies to enhance user safety and clarity, driven by global regulations like the U.K.’s Online Safety Act and the EU’s Digital Services Act. These changes aim to foster a more respectful community, addressing past issues of self-seriousness and lack of humor. Key updates include a more detailed complaints and appeals process, with an informal dispute resolution option. The revised guidelines emphasize safety, respect, authenticity, and rule-following, covering areas like hate speech, harassment, and harmful content. While Bluesky aims to provide users with community-building tools, there's a preference for the app to handle moderation. The Privacy and Copyright Policies are also being updated to comply with global laws.

Cybersecurity

Russian Hackers Breached Norwegian Dam, Triggering Water Release

In early April, Russian hackers briefly seized control of the Bremanger dam in western Norway, releasing millions of gallons of water before the attack was halted.

The cyberattack involved opening a floodgate, resulting in the discharge of approximately three Olympic-sized swimming pools worth of water over a four-hour period.

Beate Gangås, the head of Norway’s security police service, attributed the attack to Russian hackers.

The Russian embassy has denied any involvement in the incident.

This event follows a pattern of alleged Russian cyberattacks targeting Western energy systems, including previous attacks on Ukraine's power grid in 2015 and 2016.

Business

Kodak Refutes Shutdown Rumors, Cites Debt Management Plans

Eastman Kodak has refuted media reports suggesting the company is shutting down due to financial difficulties.

Recent reports highlighted Kodak's financial challenges and concerns about meeting debt obligations.

Kodak issued a press release stating it has no plans to cease operations or file for bankruptcy.

The company plans to repay, extend, or refinance its debt and expects a stronger balance sheet next year.

Kodak intends to use $300 million from its pension plan termination to address a portion of its $477 million term debt.

Despite past financial struggles, including a 2012 bankruptcy, Kodak aims to navigate its current challenges.

Artificial Intelligence

Multiverse Computing Unveils Tiny, High-Performing AI Models for IoT Devices

European AI startup Multiverse Computing has launched SuperFly and ChickBrain, two highly compressed AI models designed for use in IoT devices, smartphones, and PCs. SuperFly, with 94 million parameters, is designed for devices with limited data, enabling voice command operation of appliances. ChickBrain, at 3.2 billion parameters, offers reasoning capabilities and outperforms the original Meta's Llama 3.1 8B model in benchmarks. Multiverse is in discussions with major device manufacturers like Apple, Samsung, Sony, and HP, and offers its compression tech via an API on AWS.