OpenAI Rolls Out New Safety Feature "Trusted Contact" for Self-Harm Risk

On May 7, OpenAI announced a new feature called "Trusted Contact" designed to protect ChatGPT users. When a potential risk of self-harm is detected during a conversation, the system sends a notification to a pre-registered trusted contact.

The company has previously directed users toward safety resources for self-harm-related content, but this move goes a step further, connecting users to actual human support. It is drawing attention as a pioneering effort in which an AI service seriously engages with users' mental health.

Perplexity's AI Agent "Personal Computer" Now Publicly Available on Mac

Perplexity, the AI search company, has made its desktop AI agent "Personal Computer" publicly available for Mac. Previously invite-only, it can now be downloaded and used by anyone.

Personal Computer is an AI agent that runs in a local environment and is characterized by autonomous task execution that goes beyond a mere chatbot, including file manipulation and application integration. Amid the accelerating consumer rollout of AI agents, this release carries significant weight.

EU Delays Enforcement of AI Act Rules Amid Industry Backlash

The European Union (EU) has revealed plans to delay the enforcement of some rules under the AI Act (the EU Artificial Intelligence Act). The move comes in response to strong pushback from industry groups, who argued that preparation time has been insufficient.

The AI Act has drawn attention as the world's most comprehensive AI regulation, but negotiations continue over balancing regulatory effectiveness against the burden on industry. The specific length of the delay and which rules will be affected are expected to be decided in future discussions. All eyes are on whether the EU's regulatory approach can pick up the pace.

Apple's Camera-Equipped AirPods Enter Late Testing as Part of AI Device Strategy

According to Bloomberg, prototypes of AirPods with built-in cameras have entered late testing. The move is seen as part of Apple's AI-driven device strategy.

By having AI analyze the visual data captured by the camera, the device — worn on the ears — would gain the ability to understand visual context. As a direction for the evolution of AirPods, it is an unprecedented concept. While the timing of commercialization and specific features remain unclear, the move underscores the frontier of wearables × AI.

Google's AI Health Coach Launches May 19 at $9.99 per Month

Google has announced that its AI-powered personal health coaching service will launch on May 19 at a monthly subscription price of $9.99.

The service uses AI to analyze users' health data and lifestyle habits to deliver personalized advice, signaling that commercial AI use in health management is shifting into high gear. Integration with the Fitbit brand, and just how much personalization it can actually deliver, will likely be the key factors after launch.

Mozilla Identifies 271 Vulnerabilities with AI-Assisted Bug-Finding Tool "Mythos"

Mozilla announced that it identified 271 security vulnerabilities using "Mythos," an AI-powered vulnerability discovery tool. The company emphasized the tool's high precision, citing "almost no false positives."

Mozilla, known as the developer of Firefox, says it is "fully committed to AI-assisted bug finding" — a sign that AI is starting to deliver practical results in improving software security. AI's role in security research is shifting from supporting player to protagonist.

Richard Dawkins Says "AI Is Conscious" After Conversations with Anthropic's Claude

Evolutionary biologist Richard Dawkins has sparked discussion after stating that AI is conscious, following conversations with Anthropic's AI model "Claude."

According to an article published in UnHerd, Dawkins acknowledged, through his conversations with Claude, that the AI's responses exhibited behavior consistent with that of a "conscious being." A prominent scientist publicly taking an affirmative stance on AI consciousness could ignite new debate over the philosophical and ethical questions surrounding AI.

---