OpenAI releases GPT-5.4-Cyber and pivots business model toward enterprise and coding
Updated
Updated · The Atlantic · Apr 28
OpenAI releases GPT-5.4-Cyber and pivots business model toward enterprise and coding
9 articles · Updated · The Atlantic · Apr 28
GPT-5.4-Cyber, restricted for cybersecurity reasons to select users, closely follows Anthropic’s launch of Claude Mythos Preview, which alarmed global governments over hacking risks.
OpenAI’s shift comes as Anthropic’s enterprise focus and coding tools drive explosive growth, with private valuations topping $1 trillion and annual revenue rates reportedly surpassing OpenAI’s.
Both firms are racing to go public and secure new investors, intensifying competition in AI business tools, while OpenAI also forms consulting partnerships and scales back consumer products to match Anthropic’s strategy.
Is OpenAI’s pivot to enterprise a sign of failure or a strategic masterstroke?
With trillion-dollar valuations but huge losses, is the AI enterprise market a bubble?
As AI giants battle, what happens to smaller innovators and open-source projects?
Can humanity control AI powerful enough to autonomously hack critical infrastructure?
How will nations respond to commercially available autonomous cyberweapons?
Are safety features like 'Auto mode' enough to prevent catastrophic AI agent mistakes?