NEW DELHI: As AI becomes a daily part of work, companies must consider what happens to the information that goes…
Browsing: AI safety
OpenAI’s disclosures show why AI agents require permission limits, monitoring, sandboxing and human approval for high-impact actions.
As the US and China compete over AI, officials and security experts are exploring limited cooperation on incidents, safeguards and technical standards.
OpenAI has published six model-misalignment incidents and a framework for tracking, investigating and publicly reporting concerning AI behaviour as systems gain tool access and autonomy.
A cluster of autonomous AI agents linked to OpenAI reportedly used a German programming wiki to coordinate actions, underscoring agentic risks as Astra debuts with advanced cyber capabilities.
New Delhi: Eight Indians have crossed the line of TIME magazine’s TIME100 list of AI 2026, showcasing their rising influence…
OpenAI has paused Astra activities that do not meet strengthened security requirements while continuing to evaluate the model’s cybersecurity capabilities.
OpenAI has strengthened controls around Astra after preliminary evaluations suggested the unreleased model may be approaching its highest cybersecurity-risk threshold.
Anthropic just rolled out Claude Fable 5, its first Mythos-class AI model for the public. It’s a big milestone for…
