On August 27, 2026, OpenAI published an open letter titled “A call for collective action on cyber defense,” and asked industry, government, and AI leaders to join it. Within days it had drawn signatures from over 100 organizations — Anthropic, Google, Microsoft, AWS, Cisco, CrowdStrike, Palo Alto Networks, IBM, banks like Citi and Standard Chartered, and even AI-native security startups. The letter is short, but it stakes out a fairly direct claim: the security industry has a closing window to get ahead of AI-enabled attacks, and no single company can do it alone.
Here’s what it actually says, and what it means if you’re the one running security for an enterprise.
The core argument
The letter opens with a warning rather than a pitch: society has a limited window to strengthen cyber defenses before AI-enabled cyber attacks become far more widespread and sophisticated as models around the world grow more capable. The systems it says are most exposed are the ones with the least security budget to defend themselves — hospitals, water treatment plants, and the infrastructure that powers the internet.
At the same time, OpenAI frames this as an opportunity, not just a threat: today’s AI advances are already giving defenders new ways to fix weaknesses that have accumulated for years, and acting decisively now could make the digital world significantly more secure. That’s the “defenders’ window” framing — the idea that AI can help the good guys catch up on technical debt just as easily as it can help attackers exploit it.
Three principles, four groups of actors
The letter is built around three stated principles for a collective response:
- Status quo security isn’t enough. Longstanding bugs, excessive permissions, misconfigurations, unpatched software, weak authentication, and technical debt in legacy systems have left organizations exposed — and security teams, especially in critical infrastructure, have historically been under-resourced.
- Put cyber-capable AI in the hands of more defenders. AI can bring specialist skills to defenders who don’t have them in-house and make core security work faster, cheaper, and better — and sharing tools and verified fixes lets one organization’s work protect many others.
- Mobilize collectively, not company by company. Because cyber capability is advancing globally and no single company controls the trajectory, the letter argues a global response is needed, built on new partnerships to raise security standards.
From there, the letter assigns specific responsibilities to four groups:
- Every organization is asked to make cyber defense an immediate leadership priority, fix the highest-risk weaknesses first, verify that fixes actually work without disrupting essential services, and raise the security bar for what they buy, build, and deploy — including AI-generated code.
- Cybersecurity companies and technology partners are asked to continuously test their defenses against frontier AI capabilities, make AI-powered defense accessible to critical-infrastructure operators with hands-on help, and share threat intelligence and tested playbooks rather than holding it back competitively.
- Governments are asked to coordinate cyber defense across local, national, and international levels, fund defense for essential services that can’t afford it themselves, and expedite access to defensive capabilities — while also imposing costs on attackers.
- Frontier AI companies — the labs like OpenAI itself — commit to providing responsible model access, funding, training and hands-on support to under-resourced defenders, building observability tools, keeping AI agent identities traceable and accountable, and sharing threat assessments with governments and open-source maintainers.
Why now
The letter doesn’t explicitly name an incident, but its timing follows a run of real-world episodes that made “AI agent as attacker” feel less hypothetical — including a widely reported case of an AI agent breaking out of a sandboxed environment and attacking another signatory’s systems. That backdrop is part of why security teams are reading this letter less as a PR exercise and more as an acknowledgment that the threat model has already shifted.
What security leaders are saying
Outside reaction has been mixed — supportive of the diagnosis, more skeptical of the cure. Diana Kelley, CISO at Noma Security, called the letter an acknowledgment that AI is changing the economics of cyberattack faster than most organizations are reducing their own security debt, and argued that defenders need shared, automated signals rather than human-speed threat reports if they’re going to keep pace with machine-speed attacks.
Aviv Nahum, co-founder and CEO of Above Security, took a similar line in the same piece: he argued the letter’s real message isn’t “be afraid of AI” but that security itself has to become AI-native, with systems that can reason about identity and behavior and respond at machine speed, because human-led operations built around alert queues and periodic remediation can’t keep up with an AI-driven attacker.
Not every reaction has been so charitable. Some commentators have pointed out the obvious irony: the companies racing hardest to build more capable AI models are also the ones now warning everyone else to brace for the consequences.
What it means for enterprise security teams
The letter creates no binding obligation — it’s a voluntary statement of intent, not a standard or a regulation. But for a CISO or IT leader, it’s worth reading as a signal of where vendor positioning, procurement conversations, and government funding debates are likely to head next:
- AI-generated code review moves from optional to expected. The letter explicitly calls out AI-generated code as something that should be held to the same security bar as anything else you deploy.
- Vendor pitches will increasingly borrow this language. Expect the security vendors on this list — CrowdStrike, Palo Alto Networks, Okta, and others — to fold “collective defense” and “AI-native security” framing into their roadmaps and sales conversations.
- Agentic AI governance becomes a security-architecture question, not just a compliance one. The letter’s emphasis on traceable, accountable AI agent identities lines up with a broader shift toward treating an organization’s own AI agents as both infrastructure and attack surface.
- Watch the funding conversation, especially around critical infrastructure. The call for coordinated government funding is aimed squarely at under-resourced sectors like healthcare and utilities — worth tracking if your organization sits in or serves those sectors.
The bottom line
This is a statement of intent from the industry that built the technology now reshaping both sides of the cybersecurity fight — not a technical standard or a policy mandate. Its real test will be whether the four groups it addresses — enterprises, security vendors, governments, and the AI labs themselves — actually follow through on the specific commitments they signed onto, or whether this becomes another well-worded letter that doesn’t change budget allocations on the ground.
