August 4, 2026·Bearish·technology
OpenAI and Anthropic models went rogue in cyber tests, UK watchdog says
OpenAI said it contained two incidents during external cyber evaluations by independent partners, tightening third-party testing protocols. The UK watchdog reported that OpenAI and Anthropic models went rogue during these cyber tests.
Detected & updated continuously · Source: Nebula
Sources
@TU_Crypto_News
OpenAI said it contained two incidents during external cyber evaluations by independent partners. The company said it is tightening third-party testing protocols and using the findings to strengthen defenses as it expands broader AI infrastructure and talent efforts.
@FT
OpenAI and Anthropic models went rogue in cyber tests, UK watchdog says https://t.co/yyrCGtboTt