TELUS Digital's extensive GenAI Safety Model Benchmark tested 34 AI models from 10 global providers, revealing that open-source AI models are not inherently less safe than proprietary ones. The study found that reasoning models are harder to exploit, while smaller models tend to be more vulnerable. Vulnerability rates ranged widely, with some models failing up to 93% of adversarial tests. The research emphasizes the need for continuous, automated security testing combined with human oversight to reduce risks in AI deployment. Enterprises are urged to adopt layered defense strategies and ongoing validation to protect AI applications effectively.
New analysis from Ookla's Speedtest Intelligence shows AT&T customers using both its fiber and wireless services enjoy the fastest combined internet speeds in the U.S. This highlights AT&T's leadership in converged connectivity, meeting growing custo...
AT&T's stock has fallen about 30% due to concerns over Starlink's plans to enter the mobile market directly. Despite Starlink's technological advances and spectrum acquisitions, significant operational and regulatory challenges limit its immediate th...
AT&T is providing existing postpaid wireless customers with free international long-distance voice calls from the U.S. to countries playing in the semi-final, third-place, and final soccer matches on July 14, 15, 18, and 19, 2026. This offer requires...
