OpenAI locks GPT-5.5-Cyber behind velvet rope despite slamming Anthropic for doing exactly that

OpenAI has announced a limited-access rollout of its new GPT-5.5-Cyber model, restricting availability to a curated group of "cyber defenders." This move represents a significant shift in the company's public positioning, coming mere weeks after OpenAI criticized competitors for employing similar gatekeeping strategies. The decision has sparked debate within the AI industry about transparency, access equity, and the consistency of major AI companies' stated values.

OpenAI's GPT-5.5-Cyber model will debut through a restricted beta program rather than broader public access. The company is limiting initial availability to vetted cybersecurity professionals and researchers it deems appropriate users. This controlled rollout contrasts sharply with OpenAI's recent public statements criticizing Anthropic for implementing comparable access restrictions on their advanced models. The timing of these opposing positions has raised questions about the authenticity of OpenAI's earlier advocacy for open AI development.

• Credibility concerns: OpenAI's shift undermines its recent messaging about responsible AI democratization and may damage trust with stakeholders
• Competitive inconsistency: Both leading AI labs now employ restricted access models, despite public criticism of the practice
• Security versus access debate: Limiting cybersecurity AI tools to approved users may reflect legitimate safety concerns but also creates barriers to broader security research
• Market concentration: Restricted access policies may entrench advantages for well-connected researchers and institutions
• Industry precedent: The move signals that premier AI capabilities increasingly launch through controlled distribution regardless of public positioning

The GPT-5.5-Cyber controversy highlights fundamental tensions in AI development between innovation speed, safety considerations, and equitable access. As AI systems become more capable and potentially consequential, companies face genuine security dilemmas about distribution. However, OpenAI's recent criticism of identical practices creates credibility questions that extend beyond this single product launch. The episode demonstrates how competitive pressures and safety concerns may ultimately drive similar decisions across the industry, regardless of public rhetoric. For stakeholders—from researchers to policymakers—this inconsistency complicates efforts to establish consistent governance frameworks for advanced AI systems.