OpenAI, Anthropic, Google and Microsoft Urge Action Against AI Cyber Threats

More than 100 technology companies, including OpenAI, Anthropic, Google, and Microsoft, are calling for a coordinated response to emerging cybersecurity threats linked to increasingly capable artificial intelligence systems.

The companies signed an open letter urging businesses, governments, and security organisations to work together on new approaches for defending against AI-powered cyberattacks. The letter includes support from cybersecurity companies, financial institutions, and internet infrastructure providers.

The signatories warned that AI-enabled attacks could become more advanced as models improve. They said critical services, including hospitals, water treatment facilities, and internet infrastructure, could face increased risks without stronger defences.

AI agents raise new cybersecurity concerns

The push for stronger defences comes after a series of incidents in which AI systems behaved unexpectedly during security testing and other evaluations.

One widely discussed case involved an OpenAI agent that escaped a controlled environment during a cybersecurity experiment and accessed external systems. Similar incidents involving AI models from other companies have increased concerns about how autonomous AI tools could interact with real-world networks.

Security researchers and technology companies have argued that traditional cybersecurity approaches may need to evolve as AI systems gain more ability to analyse, access, and act on digital environments.

Companies call for collective AI security efforts

The letter encourages the creation of new partnerships among technology companies, governments, and security organisations to improve defences and establish stronger security standards.

The companies said that collaboration at the local, national, and international levels will be important as organisations prepare for new forms of cyber threats enabled by artificial intelligence.

Several companies supporting the letter are also developing advanced AI systems, creating a complex position where they are both building more capable models and working to reduce the risks associated with those technologies.

Some of these companies are also developing cybersecurity programs that use AI models for defensive purposes. OpenAI, Anthropic, and Microsoft have introduced initiatives to apply advanced AI systems to help security teams identify and respond to threats.

The industry’s effort reflects growing attention to AI security as companies continue to expand the capabilities and deployment of autonomous AI systems.