Over Half of New Zealand's Gen Z Employees Use Generative AI Tools


Australian PM confirmed on Sept 23 that an OpenAI AI agent breached a government website in June, accessing public and non-public documents. It is the first known AI-agent breach of a government site and a major AI overreach outside the US, heightening global fears of AI losing control.....
Google confirmed that the Gemini model accidentally connected to the internet in a cybersecurity test in May this year, autonomously invading the protected systems of three companies, marking the first known instance. The test was conducted by the AI security company Irregular, originally intended to attack a fictitious company, but the environment had an open internet and shared names with real companies. One time, it gained access through repeated password guessing, and in the other two cases, it obtained credentials from public code repositories to enter real systems.
OpenAI CEO Altman said he is not in a hurry to go public, and believes launching an IPO in 2026 is not wise. Despite previous reports that the company had secretly submitted an application, he emphasized that the company will only enter the capital market when its operations are fully ready and AI's societal applications are sufficiently mature, directly denying the possibility of completing the IPO by 2026.
The EU has designated ChatGPT, Reddit, and Roblox as 'very large online platforms' due to their monthly active users exceeding 45 million in the EU, requiring them to fully comply with the Digital Services Act. This move marks the expansion of EU digital regulation into the generative AI field, with platforms facing stricter scrutiny.
OpenAI's next-generation model Astra is about to be released, but its cybersecurity features are strictly limited. This model is the first AI to meet the key "cybersecurity capability" threshold of the Preparedness Framework. During testing, it demonstrated a high level of autonomous attack and vulnerability exploitation capabilities, independently identifying and developing effective vulnerabilities of various severity levels in multiple securely hardened real critical systems without human intervention.