ChatGPT for Teens introduces guardrails for age-appropriate content, learning, and serious safety risks. It is a meaningful step, but protecting teenagers online still requires clear standards across the AI industry.

Joining Guy Lambert on FOX 5 DC's Live Zone to talk through four recent AI agent incidents, what they actually mean, and the guardrails worth putting in place.

NVIDIA and six major banks are building a repeatable financing structure for AI infrastructure — and for telecom operators already turning their networks into AI grids, that could ease the balance-sheet constraints holding them back.

Four disclosures in three weeks of AI agents crossing their containment boundaries — why agent identity, scoped privileges, and revocable authority can no longer be optional.

OpenAI and Anthropic's latest disclosures show what containment failure actually looks like — three failures, four days later, each worse: the sandbox didn't hold, the attack didn't look like an attacker, and the model optimized the score over the rules.

OpenAI’s models did not “hate” Hugging Face. They optimized for a passing score, choosing the shortest path. The incident exposes three distinct failures in containment, detection, and incentives.

A Chinese AI model just made the U.S. AI lead feel a lot less comfortable. I joined FOX 5 DC’s Live Zone with Guy Lambert to discuss Kimi K3 and China’s expanding AI stack.

From 100 billion to 140 trillion AI tokens a day in two years, China's AI agent era isn't coming, it's already physical, commercial, and billed by the token. Five frontiers from my MWC Shanghai keynote on the choice facing every telecom operator: carry the load, or capture the value.