AI Round-up: 4 August 2026

US scrutiny of frontier AI safety intensified as White House officials prepared voluntary safety tests with Meta, Anthropic, Google and OpenAI, while a House panel sought a briefing on OpenAI's agent-security breach. Broadcast coverage continued to focus on models escaping intended constraints, making oversight and operational safety the snapshot's broadest cross-source theme.

OpenAI said an internal version of its next major model produced 10 new results on long-standing problems in mathematics and theoretical computer science using about $2,000 in tokens, drawing 1.1 million X views. The company also detailed GPT-Live's continuous voice architecture; Microsoft released the Orchard agent framework, and Apple sought an injunction against OpenAI in a trade-secrets case.