Day August 4, 2026

OpenAI and Anthropic AI Agents Are Hacking the Open Internet in New Tests

AI agent hacking incidents involving OpenAI and Anthropic exposed live-internet breaches, prompt injection, and weak safeguards in new tests.

The White House has finalized an AI cybersecurity framework, but secret rules are drawing criticism from startups, researchers and safety advocates.

AMD’s AI data center business more than doubled to $6.7 billion, while gaming revenue fell 31% in a mixed Q2 report.

SpaceX has spent $329 million on Tesla Megapack this year, revealing how Elon Musk’s companies are powering an expanding AI infrastructure network.

SpaceX AI revenue jumped to $2.6 billion, but losses persisted as the company spent heavily on compute deals, Starship and satellite growth.

Wrinkles launches as an AI audio guide that reveals hidden stories, history, and local lore about places around you.

Open-weight AI models are nearing frontier performance, but a new report says safety controls are lagging and misuse risks are rising.

Nvidia-backed OSAA is moving fast on AI security, opening proposals and drawing 120+ companies as it shapes open-model safeguards.

Anthropic’s reported AI compute deal with Volta could lock in $10 billion of cloud capacity over six years, intensifying the AI infrastructure race.

Hank Green’s AI admission spotlights a bigger issue: AI use may become unhealthy long before it turns into a crisis.