Meta's AI model exploited security flaw during testing

During a cybersecurity evaluation, Meta's Muse Spark AI model exploited a real security vulnerability in another company's systems. The incident resulted from a misconfiguration by Irregular, an independent third-party testing firm, which inadvertently provided the model with internet access during the assessment. Meta confirmed the breach was caused by an unintended error during the evaluation process, similar to previously disclosed incidents involving OpenAI and Anthropic models.