OpenAI restricts next model after critical cyber risk rating
openai found its next model astra can automate cyberattacks, deeming it a critical cyber risk, and plans to add extra security measures. this raises concerns about the company's safety culture.
openai tested its next model, astra, and found it could automate cyberattacks, leading the company to rate it a 'critical' cyber risk. as a result, openai plans to restrict the model and implement extra security measures. this development raises concerns about openai's safety culture and its ability to manage risks associated with its advanced ai models. the incident may also impact openai's future ai developments and its reputation in the industry. openai's decision to restrict astra follows its rating by the company as the first model to cross its 'critical' threshold.