OpenAI is extending a hand to the European Union, granting the bloc access to its newly developed cyber defense AI model, GPT-5.5-Cyber. This move signals a commitment to transparency and collaborative security efforts within the burgeoning AI landscape. Meanwhile, a key competitor, Anthropic, remains more guarded with its own sophisticated cyber AI, Mythos, yet to be shared with EU entities.
The collaboration with OpenAI will involve a diverse range of European partners, including businesses, governmental bodies, cybersecurity agencies, and EU institutions like the EU AI Office. These stakeholders will gain access to GPT-5.5-Cyber, a specialized iteration of OpenAI’s latest AI technology, designed to bolster cyber defenses. OpenAI had previously announced a limited preview of this model to a select group of vetted cybersecurity teams, a rollout that closely followed Anthropic’s release of Mythos. The introduction of Mythos had, at the time, sparked considerable concern regarding its potential misuse in cyberattacks targeting critical software infrastructure.
The European Commission has welcomed OpenAI’s proactive stance. A spokesperson for the Commission, Thomas Regnier, stated, “We welcome OpenAI’s transparency and intent to give commission access to new model.” He confirmed ongoing discussions with OpenAI, with further dialogues scheduled this week to finalize the terms of access. “This will allow us to follow deployment of the model very closely, and address security concerns,” Regnier elaborated, highlighting the EU’s commitment to rigorous oversight.
In contrast, Anthropic’s approach to sharing Mythos with the EU appears to be on a different trajectory. Despite releasing its model a month prior, the company has yet to grant the EU preview access for review. Regnier indicated that while discussions with Anthropic are ongoing, they are “at a different stage” compared to the agreements being forged with OpenAI. He noted that while the Commission has engaged in several meetings with Anthropic, the conversations “are not yet at the same stage as the solution we have on the table from OpenAI.”
This divergence in strategy between OpenAI and Anthropic underscores the complex dynamics at play in the race to develop and deploy advanced AI for cybersecurity. OpenAI’s approach appears to emphasize broad collaboration and democratizing access to defensive tools, framed as an “OpenAI EU Cyber Action Plan.” George Osborne, Head of OpenAI for Countries, articulated this vision: “AI labs like ours shouldn’t be the sole arbiters of cyber safety as resilience depends on trusted partners working together.” He added, “The latest cyber AI capabilities should be available for Europe’s many defenders, not just the few, and we want to help make that happen.” The plan aims to work with European policymakers, institutions, and businesses by providing access to defensive tools that trusted actors can leverage to enhance collective security, support public safety, and align with European priorities.
The differing strategies raise important questions about AI governance and the responsible deployment of powerful technologies. While Anthropic’s reticence might stem from a desire for more stringent control or a belief in a phased release strategy, OpenAI’s open invitation suggests a belief that wider access, coupled with collaboration, is key to building robust cyber resilience. The EU’s engagement with both companies will be closely watched as it navigates the complex terrain of AI regulation and its implications for national and international security.
Anthropic has been approached for comment.
Original article, Author: Tobias. If you wish to reprint this article, please indicate the source:https://aicnbc.com/21585.html
