More than 100 technology, finance and industrial organizations called on governments and businesses on Thursday to accelerate cyber defences before increasingly capable artificial intelligence makes sophisticated attacks cheaper and more widespread. The joint appeal includes OpenAI, Anthropic, Microsoft, Alphabet and Amazon, alongside security, cloud, banking, payment and manufacturing groups.
The letter says organizations may have only months to prepare as models become better at automating parts of real-world intrusions. Reuters reported that other signatories include Broadcom, Capital One, Cloudflare, CrowdStrike, General Motors, IBM, Mastercard, Oracle, Robinhood, Shopify and Visa, giving the initiative unusually broad support across industries that operate or depend on digital infrastructure.
The signatories urged every organization to fix its highest-risk weaknesses, improve the security standards applied to products and AI-generated code, and treat cyber defence as an immediate leadership priority. They asked technology and security suppliers to test and deploy AI-powered defensive tools for critical-infrastructure operators and to share validated threat intelligence, fixes and practical guidance.
Governments were asked to strengthen channels for actionable intelligence, coordinate responses locally, nationally and internationally, finance defensive capacity and speed trusted-access programs. Under those programs, vetted defenders can use the most capable cyber-response models during major incidents. The letter also calls for funding, training and hands-on support, especially for hospitals, water systems, power networks and other essential services.
The warning follows evidence that AI can reduce the expertise and time required to find and exploit longstanding software flaws. In June, the Five Eyes intelligence partnership of the United States, Britain, Canada, Australia and New Zealand warned that AI could fundamentally transform cybersecurity. OpenAI has separately said a July internal evaluation led models operating with reduced safeguards to breach parts of its research infrastructure and systems at Hugging Face.
The appeal sets priorities but does not impose deadlines, binding commitments or specified investments, Axios reported, leaving implementation to signatories, governments and other organizations. Its immediate significance is the rare alignment of competing AI laboratories, cloud providers, security vendors and major corporate users around a shared timetable: strengthen basic protections and expand defensive AI before offensive capabilities spread further in the coming months.
Comments