How Dan O'Dowd Designs Software That "Never Fails"—and Why It Matters

AI-generated, human-reviewed.

On this episode of Intelligent Machines, cybersecurity pioneer Dan O'Dowd warns that insecure software in power grids, hospitals, pipelines, and self-driving cars could lead to catastrophic failures—even mass starvation in the event of a prolonged outage. O'Dowd, founder of The Dawn Project and a leader in ultra-reliable operating systems, argues that the industry’s focus on speed and cost has left essential systems dangerously exposed to both bugs and cyberattacks.

How Dan O’Dowd Builds “Unhackable” Software for High-Stakes Systems

Dan O'Dowd’s credentials stem from decades spent engineering the backbone software for nuclear bombers, NASA missions, and critical aircraft. On Intelligent Machines, he detailed a meticulous methodology: investing thousands of dollars per line of code, banning programming practices known to cause bugs, and subjecting every component to a rigorous, multi-person review before deployment.

His team’s operating system, Integrity, is used in the most sensitive military and aviation hardware—where failure or hacks are unacceptable. O'Dowd emphasized that the software underpinning such systems has received the NSA’s highest security certification and has not been breached despite open challenges to hackers.

Why Most Critical Infrastructure Remains Vulnerable

Despite consensus from top security officials at the NSA and FBI that the power grid and other utilities are under constant threat, O’Dowd explained that operators are reluctant to upgrade due to enormous costs and regulatory inertia. The existing infrastructure relies heavily on commercial software designed for convenience and speed, not for safety or reliability under attack.

According to O'Dowd, complacency and the “checkbox mentality”—doing only the minimum to meet regulatory requirements—means critical vulnerabilities are left unaddressed. He cited both software errors (bugs) and external hacking as risks, but stressed that robust software can compensate for most hardware weaknesses.

The Perils of “Move Fast and Break Things” in Safety-Critical Applications

Drawing a stark contrast with consumer tech, O’Dowd criticized Silicon Valley’s ethos of “move fast and break things,” noting that cutting corners and prioritizing rapid feature release is disastrous when lives depend on flawless operation. He highlighted how much commercial software is cobbled together from third-party libraries, each introducing new vulnerabilities—meaning even well-intentioned engineers may inadvertently import thousands of flaws into supposedly secure systems.

O’Dowd’s warning: in essentials like the electric grid or autonomous vehicles, there is no room for error—the consequences of a single bug or exploit can be deadly.

Self-Driving Cars: Why O’Dowd Says Most Aren’t Safe

O’Dowd was unsparing in his critique of current self-driving technology, especially Tesla’s rollout of “beta” features to ordinary users. He explained that systems like Waymo, which underwent years of controlled testing, are far safer, whereas Tesla’s software has been documented to miss critical cues like stop signs or school buses. Even beyond coding bugs, O’Dowd called out hardware design flaws—like camera placement that makes it impossible for the car to see necessary angles at intersections.

The combination of over-the-air updates and identical software across cars only amplifies the risk: a single exploit could affect millions of vehicles simultaneously.

What You Need to Know

• Critical infrastructure (power grids, hospitals, pipelines, etc.) is at growing risk due to insecure and outdated software.
• Most software in these systems was not built to military-grade safety standards and contains thousands of vulnerabilities.
• Upgrading infrastructure software is expensive, leading to resistance from operators and slow regulatory change.
• Dan O’Dowd advocates for military-style development: careful review, banning dangerous programming habits, and relentless testing.
• Consumer "move fast and break things" methods are disastrously inappropriate for life-critical systems.
• Self-driving car safety is not just a coding problem; hardware design (like camera placement) can undermine reliability, and broad software updates heighten systemic risk.
• The public is often unaware of these risks unless high-profile failures occur—O’Dowd’s projects aim to change that.

The Bottom Line

According to Dan O’Dowd on Intelligent Machines, the greatest cybersecurity and public safety threats come not just from hackers, but from a widespread culture of software shortcuts and regulatory complacency in America’s most critical infrastructure. True safety requires a relentless commitment to software that never fails—an ethos rarely found outside military and aerospace domains. As threats mount, meaningful reform and public pressure will be needed to spur costly but necessary upgrades before disaster strikes.

Listen and subscribe to Intelligent Machines for expert analysis on technology, AI, and the future of secure systems:
https://twit.tv/shows/intelligent-machines/episodes/884