OpenAI halts GPT-6.1 Astra launch after researchers raised concerns
Big week in tech safety: OpenAI has put the brakes on its GPT-6.1 Astra launch owing to safety concerns raised by its researchers during internal testing.
According to OpenAI's head of safety systems, Saachi Jain, even though the new model had become more efficient in completing tasks, the company needed to balance its capability against unauthorized behavior, and the model "didn't quite meet the bar."
Cisco patch, NVIDIA platform, ChatGPT misuse
Cisco patched a serious flaw in its Catalyst SD-WAN Manager that could let hackers take over devices. Users are urged to update as soon as possible since there is no quick fix.
Meanwhile, NVIDIA rolled out its new Open Agent Safety Platform to help keep AI agents in check as they get more powerful.
And heads up: cybersecurity firm Huntress found hackers using ChatGPT's Custom GPTs feature to sneak malware onto people's computers, reminding us that cool tech always needs strong defenses.