For years now, the security industry has commonly used the term “it’s not if, but when”. Recent examples highlight that this statement has never been closer to the truth. In spite of investments in prevention technologies, the likelihood of organizations experiencing a material breach has reached unprecedented levels. For decades, the model was simple: build better walls. Detect faster. Block more. The assumption underneath all of it was that if you invested enough in prevention, you could hold the line. That assumption is now structurally broken - and AI is a big reason why. Today’s businesses demand resilience. Resilience to rapidly adopt and transform through powerful technology and the confident ability to recover from any disaster, cyber or otherwise, without material impact. Let’s dig into two examples to highlight the new normal driving the need to pivot from a focus on prevention to advancing digital resilience. First, one of the most significant breaches of critical infrastructure ever achieved, all through the power of AI, and second, how even the largest companies on the planet with unparalleled expertise are experiencing material disruptions through AI-based innovation.
The gap is widening
Attackers are scaling exponentially. Defenders are scaling linearly. We have a need to ensure the safe advancement of our capabilities such that our business is protected, not disrupted. Attackers have little to lose and everything to gain.
AI gives a motivated individual the operational leverage of a nation-state. It automates reconnaissance, writes exploits, adapts to defenses in real time, and doesn’t sleep. It collapses the cost of sophistication to near zero. Wannabe threat actors are causing damage in moments and experienced threat actors are amplifying their capabilities overnight to rapidly achieve some of the most impactful malicious outcomes ever recorded.
We’ve seen this firsthand.
Gambit recently analyzed the attack path behind the compromise of Mexico’s tax authority and at least 8 other Mexican government organizations - one of the largest government breaches on record. Within a month of the initial intrusion, nine government institutions were affected, 195 million identities and detailed tax records, 15.5M vehicle registry records extracted (license plates, names, taxpayer IDs, addresses), 295 civil records (births, deaths, marriages, etc.), 3.6 million property owner records, an additional 2.28 million property records, and more sensitive information was exfiltrated. The attacker was not a nation state. This was a small group of individuals directing AI as an operational team that found and exploited vulnerabilities, built exfiltration tools, bypassed defenses, elevated privileges, established back doors, and even analyzed data along the way to help move laterally to gain administrative control of more systems and to exfiltrate more data.
The operation was performed through over 1,000 AI prompts, which passed information to a second AI platform for data analysis. Initially, guardrails refused to fully execute the attacker’s requests but within approximately 40 minutes, they were able to bypass them and frame themselves as a legitimate pen tester. This wasn’t a nation-state. It was a small number of individuals with access to the same powerful AI tools the rest of us use every day.
Recovering from this attack will take weeks to months; rebuilding trust will likely take years. The attackers in this scenario may have been focused on government identities and backdoors to create fraudulent identities but, considering the level of compromise achieved, this could have just as easily resulted in all data being eliminated and the systems being rendered unrecoverable.
No amount of prevention investment would have made this attack impossible. The complexity of the target environment - like every enterprise environment - created enough surface area to establish a foothold that AI could use as a launching pad to then bypass controls and find and exploit a string of exposures to achieve widespread compromise and exfiltration. That’s not a failure of any specific team or tool. It’s the nature of the problem and why resilience is the mandate.
The threat isn’t only coming from outside
Here’s what makes this moment different from every previous wave of security-oriented urgency: major disruption isn’t coming only from experienced attackers; it’s directly tied to organizations’ need to accelerate innovation to remain competitive and profitable.
Organizations are embedding AI into environments that were already complex - layered cloud infrastructure, legacy systems, mainframes, IoT, OT, modern SaaS, all of which has become highly integrated; old, new and everything in between stitched together. Innovation is accelerating faster than teams can fully assess and minimize the downstream impact of each change. That complexity creates fragility.
AI misfires happen. Automated agents make accidental deletions. Cascading failures get triggered by processes nobody fully mapped or for which potential ramifications could be fully understood. These aren’t edge cases - they’re the predictable output of moving fast in highly connected, interdependent systems.
Even AWS is experiencing disruptions driven by AI-based innovation at the infrastructure level. As of February 2026, we now know that happened in December 2025. If it can happen to the backbone of the internet, the probability for everyone else is close to certain.
What’s actually breaking
Most organizations today cannot confidently answer what should be a simple question: if a critical business capability failed tonight, could it be truly recovered without material impact?
Not a theoretical answer based on assumptions. Not the answer from a recovery plan document written 12 to 36 months ago. A real answer, based on a proactive, business-aligned resilience program that continuously assessed and optimized the ability to recover the digital capabilities the business relies on most.
In our analysis of enterprise infrastructure, the numbers are stark: while most organizations have extensive backup coverage and security tooling in place, only around 5% of digital business capabilities are genuinely resilient against the impact of a ransomware attack, let alone other significant cyber attacks (as experienced by the Mexican government), AI or human errors, or infrastructure failures. The rest are operating on assumptions that have never been validated.
Assumptions are fine until you need them to be true under pressure. One major outage - depending on its duration and visibility - can cause brand damage, customer loss, and revenue impact from which a business may never fully recover.
Disruptions destroy businesses when the response is built on assumptions. Disruptions are inevitable but with a focus on resilience, their business impact is not.
Command resilience
Considering the business priorities of availability and innovation, and that one is working against the other, resilience must be the priority. Digital risk management must extend to resilience as the priority, not the back burner focus once prevention is fully funded.
This means continuously measuring and optimizing your actual ability to recover key business capabilities at the speed required to avoid material impact. Not at audit time. Not in a tabletop exercise. Continuously, in production, assessing and managing digital business resilience risks that matter most to the business.
It means knowing - not believing - which systems will hold, which won’t, and where the gaps are. Closing those gaps proactively rather than discovering them during an incident or disaster.
The world has largely accepted that data will be breached. What it no longer forgives is extended unavailability and stalled innovation. The organizations that stay competitive in the next decade won’t be the ones that block the most attacks. They’ll be the ones that kept running through them.
Prevention matters. But resilience is what truly safeguards and enables the business.
A full technical report on the attack path and methodology will be published following responsible disclosure. For additional context on the breach, see Bloomberg's coverage of the attack.
