AI Safety

Kaspersky has published verification guidelines and security recommendations for students and parents using artificial intelligence tools during the school year.

OpenAI released its official report examining the cybersecurity incident that affected Hugging Face, outlining key technical failures and new safeguards.

Anthropic is redeploying its Claude Fable 5 model globally on July 1, 2026, after the US government lifted temporary export controls.

Anthropic has suspended access to Fable 5 and Mythos 5 globally following a US government export control directive restricting foreign national access.

The US government has ordered the suspension of Anthropic Fable 5 and Mythos 5 models over national security concerns, forcing the developer to disable access.

A new Telus Digital report reveals that every major generative AI model can be coaxed into unsafe behavior, leaving enterprises underprepared.

An unreleased Anthropic model known as Claude Mythos AI surfaced online, raising cybersecurity concerns about unfiltered language models and AI data governance.

OpenAI has detailed the structure and philosophy behind its Model Spec, the formal framework governing how its AI models behave across user queries. The document covers hard rules, overridable defaults, and a chain of command for resolving instruction conflicts.

OpenAI released open source teen safety tools to help developers build safer apps for users under 18, covering harmful content, dangerous activities, and age-restricted material.

Kaspersky cybersecurity experts share practical tips for parents to make AI safer for Generation Alpha children on Safer Internet Day 2026.