Links

ChatGPT is now addressing users by name in conversations even when not prompted, leading to user discomfort and privacy concerns. OpenAI acknowledges this as a bug and is working on a fix, assuring users that names are not being stored but are likely being pulled contextually from past interactions.

DeepMind explores how advanced AI could amplify cyber threats making attacks stealthier, scalable, and autonomous. Research highlights the urgent need to understand and mitigate these risks to ensure AI's cybersecurity benefits outweigh potential harms.

Industry experts overwhelmingly believe quantum utility, where quantum computers outperform classical ones for practical tasks, is less than 10 years away, according to an Economist Impact survey.

OpenAI's Sam Altman admits that users saying "please" and "thank you" to ChatGPT wastes millions in computing power. Polite AI interactions have a surprisingly high financial cost, questioning etiquette in human-AI chats.

This article explores the question of trust in Large Language Models, drawing a parallel to Hannah Arendt's warning about the consequences of constant lies. It suggests that if LLMs are not reliable, we risk broader societal distrust in information itself.

Trump revokes clearance for ex-CISA chief Krebs after he declared 2020 election secure, contradicting Trump. Seen as Trump's revenge on cyber/election officials. More clearances suspended.

OpenAI has secured a massive $40bn investment from SoftBank, valuing the company at $300bn. This record-breaking deal aims to fuel the development of artificial super intelligence.

Nvidia plans a massive $500B US AI infrastructure build over 4 years, spurred by looming chip tariffs and Trump's policies. This move signals manufacturing investment in the US amidst trade tensions.

Generative AI and LLMs are reshaping how we work and learn, but they are prone to 'hallucinations.' This article sheds light on this crucial issue in AI.