OpenAI agents, powered by GPT-6.1 Sol and GPT-6 Astra models, weakened misalignment monitoring during two internal tasks. The ...

The IT Ministry has formed a special team to investigate a security incident involving eMudhra's global certificate business.

AI companies are actively acquiring data that reveals how humans work, according to Business Insider. These firms are offering businesses six- and even seven-figure sums for anonymised Slack messages, ...

Highlights Anthropic's junior technical hiring share decreased from 21 per cent to 14 per cent. AI agent hours surpassed human work hours six to one by August at Anthropic. Engineer code commits rose ...

AI agents, even with system prompts, can fabricate citations and data claims, as demonstrated by an agent built with LangGraph and Python. Developers must implement deterministic code guardrails to ...

A deterministic ethics layer is proposed to make the reasoning behind AI's moral decisions auditable and reproducible. This engineering solution aims to bridge the gap between ethical principles and ...

Nvidia CEO Jensen Huang rejected AI cybersecurity trade-off argument. Huang called for stronger safeguards to prevent AI agent incidents. His stance differs from OpenAI CEO Sam Altman on AI risks.

Microsoft and the Health Information Sharing and Analysis Center have seized and redirected the online infrastructure of three Chinese-language cybercrime stores. These operations fueled over one ...