Anthropic's Mythos Comes Back Online
The day's biggest thread runs through Anthropic. The US government has partially lifted its block on the company's most powerful model, Claude Mythos 5, clearing it for release to a limited circle of approved "trusted partners," per MSN. PYMNTS and Telecompaper confirm the broader picture: Anthropic now has Washington's approval to bring the Mythos system back online after a period offline. It's a notable example of national-security gatekeeping shaping who gets access to frontier AI — and how fast.
That gatekeeping instinct echoed elsewhere. Nvidia's CEO, speaking via Yahoo Finance, publicly put national security ahead of the China sales his company could be making — a striking concession of commercial ground from the chipmaker at the center of the AI boom.
Meanwhile, China answered the Mythos narrative directly. Zhipu AI — known as Z.ai — released an open-weight model, GLM-5.2, that some researchers say can match Mythos on cybersecurity tasks, according to The Verge. An open model rivaling a tightly controlled American one reframes the whole export-control debate.
OpenAI Expands on Every Front
OpenAI had a busy day. It introduced three new models — Sol, Terra, and Luna — tied to GPT-5.6, per Basic Tutorials. Separately, AI analyst Zvi Mowshowitz reports that GPT-5.6's newly surfaced safety system card appears to clear the company's internal bar, signaling a green light for a fast, unrestricted rollout.
On the commercial side, HP launched a "Frontier" strategic partnership with OpenAI, scaling up an existing relationship to deploy OpenAI technology across its business and customer products.
The Compute Crunch and Its Costs
A sobering data point ties the labs together: OpenAI, Anthropic, and xAI now consume 21% of global AI compute between them, per Crypto Briefing. Three companies, a fifth of the world's AI horsepower — a concentration with real strategic weight.
Not every bet on AI is paying off, though. Ford has rehired veteran "gray beard" engineers after an AI-led effort fell short of expectations, TechCrunch reports — a reminder that automation still leans on hard-won human experience. And at Alphabet, a strategic coding pivot at DeepMind is driving key talent to rivals, with Tech Times pegging the fallout at a $270B hit and a worrying brain drain.
Europe Wants In — and the Youngest Workers Feel the Squeeze
Europe is moving on two tracks at once: loosening its AI rulebook while courting American talent. Per Bloomberg, Austria is actively pushing to lure Anthropic onto its soil as the EU softens regulation.
Closing on a human note: a Fortune analysis finds AI may already be thinning the ranks of America's youngest workers, with entry-level hiring feeling the squeeze first. As the frontier races ahead, the labor-market bill is starting to come due.