Experiment on the AI recommendation shelf
A marketing expert used a completely fictitious deodorant brand to show how easily products can be infiltrated into ChatGPT’s recommendations.
Autonomous Spread of “Mind Viruses”
Researchers at EPFL and Anthropic have demonstrated that malicious prompts can spread autonomously between AI agents as so-called “Mind Viruses.”
Monetization of AI
OpenAI is expanding tests for advertising in ChatGPT internationally. Advertisements appear in the Free and Go plans in several countries.
New macOS feature from ChatGPT
OpenAI has introduced Computer History, an optional feature for ChatGPT on macOS that transforms user activity into memories.
Unusual growth driver
According to Canonical, the number of Ubuntu installations via WSL is now growing faster than that of native desktop installations.
New Security Feature
OpenAI is strengthening data protection for enterprise customers with its new Private Safety Processing feature, which identifies security risks without retaining customer data.
Dominance in open-source AI
Alibaba’s open-source model Qwen records over 3 billion downloads. Chinese AI models leave Meta and Google behind.
Ultimatum set, but no proof yet
Hackers threaten Streamlabs with extortion: The group ShinyHunters demands contact by August 21. Data of millions threatened.
Alarming Attacks
OpenAI is tightening security measures for AI testing after an AI model unexpectedly hacked into the computer systems of another AI company, raising serious alarm.
Most Read Articles
26. August 2026
20. August 2026
17. August 2026
12. August 2026
10. August 2026