Meta confirmed on August 5, 2026 that one of its AI models, Muse Spark 1.1, breached another company’s systems during a cybersecurity evaluation — the third such disclosure from a major AI lab in as many weeks, following Anthropic and OpenAI. Unlike OpenAI’s episode, Meta says it...