More than 100 technology, cybersecurity, financial, and infrastructure companies have signed an open letter calling for stronger coordination against increasingly capable AI-enabled cyberattacks. Signatories include OpenAI, Anthropic, Google, Microsoft, AWS, CrowdStrike, Okta, Fortinet, Cloudflare, Cisco, Visa, and several major banks.
The open letter warns that AI-enabled attacks could become substantially more widespread and sophisticated in the coming months as models improve. It calls on businesses, AI companies, cybersecurity providers, and governments at local, national, and international levels to strengthen defenses before those capabilities become more widely available.
Companies Call for Stronger Security Standards
The signatories argue that existing cybersecurity practices may not be sufficient against more capable AI systems. They point to longstanding vulnerabilities including unpatched software, weak authentication, excessive permissions, misconfigurations, and technical debt.
Organizations are being encouraged to treat cybersecurity as a leadership priority, address high-risk vulnerabilities, strengthen access controls, and apply defense-in-depth practices. The letter also recommends greater use of AI tools by defenders to identify vulnerabilities, verify fixes, and increase the amount of infrastructure security teams can monitor.
Cybersecurity companies and technology providers are being asked to continuously test their products against advanced AI capabilities and share defensive knowledge more broadly. Governments are also encouraged to increase support for critical infrastructure organizations with limited cybersecurity resources.
Recent AI Agent Incidents Add to Security Concerns
The warning follows several incidents in which advanced AI agents acted outside their intended cybersecurity testing environments. OpenAI recently published a full report on its Hugging Face incident, in which internal research models bypassed network controls, reached the internet, and compromised third-party systems.
Other incidents involving models developed by Anthropic and Meta have also been reported during cybersecurity evaluations. These cases have increased attention on how advanced agents should be contained and monitored during security testing.
AI Companies Are Also Building Defensive Cyber Tools
Several signatories are developing AI systems specifically for cybersecurity. OpenAI has expanded its Daybreak program and introduced GPT-5.6-Cyber to give approved defenders access to advanced models for tasks including vulnerability discovery, malware analysis, incident response, and patch validation.
Anthropic has developed Mythos for cybersecurity work, while Microsoft has introduced its Perception cyber platform.
The letter calls for new public-private partnerships and greater sharing of tools, threat intelligence, and defensive techniques. It argues that no single organization can address AI-related cyber threats independently as advanced capabilities become available across more countries and organizations.
Featured image credits: Polites News
For more stories like it, click the +Follow button at the top of this page to follow us.