| Mobile Web

Security-focused AI models surge from vulnerability detection to reasoning

Competition is heating up over security-focused AI models designed to find software vulnerabilities and help fix them, as malicious hackers increasingly use AI in cyber attacks. The race now includes major AI developers and enterprise vendors such as Google, Cisco and Microsoft. Google introduced Gemini 3.5 Flash Cyber, Cisco unveiled its Antares small language models, and Microsoft announced MAI-Cyber-1-Flash. Startups are also moving quickly, including Cogent Security with its VR-1 reasoning model.