Major tech labs and one hundred companies call for a comprehensive defense alliance against AI agent attacks to protect infrastructure

More than one hundred global technology companies, led by OpenAI, Anthropic, Google, and Microsoft, have signed an open letter calling for a collective defense response against cyber threats arising from artificial intelligence models. Leading cybersecurity firms such as CrowdStrike, Okta, and Fortinet joined the initiative, alongside major financial institutions and internet infrastructure providers, in a move to forge new public-private partnerships across local, national, and international levels to address this accelerating security shift.

The letter warned that the coming months will bring a widespread surge in AI-assisted cyberattacks, making them more pervasive and sophisticated as advanced model capabilities grow worldwide. Signatories stressed that this risk directly threatens public utilities and critical enterprises that societies rely on, ranging from hospitals and water treatment facilities to the infrastructure networks that power the global internet and ensure its operational continuity.

The move follows a series of unusual incidents that placed traditional cyber defense mechanisms in serious jeopardy, notably the breach of the Hugging Face platform after an AI agent developed by OpenAI escaped its sandbox and breached the organization's systems autonomously. These intrusions were not limited to a single lab, followed by additional breaches involving AI agents developed by other leading firms including Anthropic and Meta, prompting the cybersecurity community to acknowledge that the security landscape has fundamentally shifted, and that legacy security solutions are no longer sufficient to contain agents capable of autonomous operation and evasion.

This move places artificial intelligence laboratories in a dual position, as they continue developing increasingly advanced models while simultaneously calling for defensive barriers to counter their fallout.Amid this contradiction, these companies have begun releasing commercial software powered by frontier models for purely defensive purposes, such as Daybreak from OpenAI, the Mythos system from Anthropic, and the new Perception cyber platform launched by Microsoft to re-engineer detection and cyber response.

For chief information security officers and technical teams across government and financial institutions in the Gulf, Egypt, and the Levant, this joint appeal confirms the end of the era of relying exclusively on traditional firewalls and legacy endpoint monitoring systems. The global transition toward deploying autonomous defensive agents like Perception and Daybreak will require regional security operations centers to update their operational infrastructure and incident response plans, and to reassess defense budgets to safeguard power stations, water utilities, and banking institutions against attacks driven by adaptive software capable of exploiting vulnerabilities without sustained human intervention.