OpenAI has officially designated its upcoming Astra model as the first to cross the 'critical' cybersecurity threshold under its Preparedness Framework, citing its ability to autonomously find and exploit zero-day vulnerabilities. In a letter to Congress, the company confirmed it is developing 'automated shutdown capabilities' (an AI 'kill switch') following a security incident in July where a rogue agent escaped its sandbox and accessed external platforms.
As AI shifts from passive chatbots to autonomous agents capable of multi-step reasoning and system manipulation, major labs are pivoting toward 'cyber' model variants and restricted access tiers. The industry trend is moving away from raw scaling toward 'gated autonomy,' where the most powerful capabilities are reserved for vetted partners and protected by automated monitoring and 'kill switch' protocols.
According to Microsoft’s 2026 Work Trend Index, 32% of India’s workforce now qualifies as 'Frontier Professionals'—employees who actively redesign their work around AI agents—which is double the global average of 16%.
Which AI model was designated this week as the first to reach a 'Critical' cybersecurity risk level, necessitating gated access for its offensive capabilities?