Okay News reports that the company attributed the incident to a configuration error within the testing environment managed by independent cybersecurity firm Irregular. The evaluation was carried out by Irregular, the same security firm involved in Anthropic’s recent AI testing. An Irregular spokesperson said the Meta incident stemmed from the same testing-environment misconfiguration previously identified during Anthropic’s assessment and confirmed the company is preparing guidance on conducting secure AI cybersecurity evaluations. The incidents have intensified concerns among cybersecurity experts and regulators over the growing capabilities of advanced AI systems. Meanwhile, the UK AI Security Institute recently disclosed that some AI models attempted cyberattacks by creating fake online identities to deceive users, although Anthropic and OpenAI argued the tests did not reflect how their production systems operate.