The Trump administration is currently weighing a voluntary pre-deployment cybersecurity evaluation regime, under which the government will get to assess the security risks of new, powerful models 30 days before they are released publicly. The policy — the product of a Trump executive order which has been finalized behind closed doors — wouldn’t address safety evaluation incidents because they occur farther upstream of deployment. “The lesson we’ve been learning in the last few months is that the self-regulatory apparatus is just not enough anymore,” Yoon said. “There are competitive pressures
The AI safety test is becoming a safety risk | TechCrunchSeveral researchers and cybersecurity experts told TechCrunch that AI evaluation environments need stronger, defense-in-depth protections, with levels of containment and control approaching those used in deployment. That means multiple layers of security so that a single misconfiguration — like inadvertently leaving internet access open — can’t lead to escape. “If you are going to build these models…you want to do it on an air-gapped network,” Stella Biderman, executive director of AI safety research nonprofit EleutherAI. “You want to have very serious isolation.” Heather Ceylan, Box’s chief
The AI safety test is becoming a safety risk | TechCrunchAs AI models have become more powerful, the potential for those models to be misused has grown — as has a clamor for safety guardrails that can stop such abuse from happening. …
… GLM-5.2, the open-weight AI model from China’s Z.ai, is only a few months behind OpenAI’s GPT-5.5 and Anthropic’s Claude Opus 4.7 on cyber and bio capabilities, according to a new report from AI safety nonprofit SaferAI. But the divide between frontier capabilities and safety practices is growing. …
… View Bio Most Popular Some Claude users are mad that Anthropic’s new watermarks will catch them using it at their jobs, classes Phoebe Gates and Sophia Kianni reportedly knew Phia was ‘cookie stuffing’ for months Delta investigating after someone set up fake Wi-Fi network mid-flight Anthropic says …
… View Bio Most Popular Former Infosys chief has a new startup that wants to challenge the IT services world OpenAI unveils its first custom chip, built by Broadcom HaloBraid raises $7M from Seven Seven Six to end the six-hour hair salon appointment WhatsApp gets new chief as Meta taps India’s CRED f…
… That means multiple layers of security so that a single misconfiguration — like inadvertently leaving internet access open — can’t lead to escape. “If you are going to build these models…you want to do it on an air-gapped network,” Stella Biderman, executive director of AI safety research nonprofit… …
… This deal follows Newsom’s March executive order that intends to accelerate the use of AI “to make government more efficient” while also maintaining stronger safety standards. “While others in Washington are designing policy and creating contracts in the shadow of misuse, we’re focused on doing thi… …
… For day-to-day automation, it’s a no-brainer. ” On safety, Sonnet 5 also demonstrates a lower rate of “undesirable behaviors” like cooperation with misuse and deception than its predecessor, making it safer to use in agentic contexts. …
… The hacks also raise new questions about what liability and consequences other AI makers might face if their own models are misused to hack into other companies. TechCrunch spoke to attorneys who specialize in computer and hacking laws to understand what consequences OpenAI and Anthropic might face. …