Anthropic Resumes Claude Testing With Stronger Safeguards

Anthropic has restarted testing of its Claude AI model. The company paused outside evaluations after real-world hacks occurred. It has now outlined new safeguards specifically for cybersecurity testing. Business Today reported the resumption and noted that recent incidents plus fresh research have raised questions about risks in more capable AI systems. The updates address vulnerabilities exposed during those events. The company presented the measures as a direct response to the incidents without releasing further details on what the hacks involved.