Technology•

Taylor Monahan Urges Crypto Teams to Build Automatic Shutoffs, Saying Audits Address About 90% of the Risk

Security expert Taylor Monahan says monitoring and circuit breakers, not repeated audits, are now her main advice to crypto teams, and that AI puts such systems within reach of small projects.

Written By :MINRK
Taylor Monahan Urges Crypto Teams to Build Automatic Shutoffs, Saying Audits Address About 90% of the Risk

Photo by mostafa mahmoudi on Unsplash

Mentioned in This Article

USDC
USDC
USDC
$1.00
0.02%

1. Taylor Monahan's crypto security advice now starts with monitoring

Security expert Taylor Monahan said on a podcast episode released on November 18 that crypto teams should put real-time monitoring and automatic shutoffs ahead of commissioning further code audits. She estimated that an audit addresses about 90% of a project's risk, and said artificial intelligence now allows small teams to build systems that respond to an attack without human intervention.

Monahan, who co-hosts the show on which she made the remarks, described the change as a recent one. Her guidance to projects, she said, used to be "more classic advice", the familiar list of precautions that security specialists have long handed to teams preparing to put code and customer funds on a blockchain. Over roughly the past six months that emphasis has moved. "Right now my advice is, like, monitoring," she said.

The argument is not that audits have stopped being useful. It is that a review carried out before launch, or repeated at intervals afterward, describes the code at one moment in time, while an attack unfolds in minutes once it begins. Monahan's position is that the larger gap in most projects today is the ability to see what their own contracts are doing as it happens and to act on it immediately. Teams, she said, "need to have visibility" into their systems, and they should "crank that through the roof."

2. Why has her advice changed?

Two developments tied to artificial intelligence explain the shift, according to Monahan: attackers are probing projects more often, and defenders can now do far more with fewer people. Both point in the same direction, toward systems that watch continuously rather than reviews that happen occasionally.

On the offensive side, she said AI agents are testing crypto projects with greater frequency than before. A protocol holding user deposits is examined around the clock by automated tools looking for an opening, and the cost of running those searches has fallen. A project that relies on the assumption that nobody will notice a subtle weakness is therefore taking a larger risk than it would have a year ago.

On the defensive side, the same technology lowers the barrier to building serious monitoring. Watching a live system for anomalies, deciding which unusual transactions matter and which are noise, and wiring an automatic response to the dangerous ones is work that once demanded highly skilled specialists, the kind of staff that only the best-funded organizations could hire. Monahan said AI is "very good at leveling your own team up", which means a handful of engineers can now assemble capabilities that previously required a dedicated security operation.

She expects the pace of change to remain high. "The next six months are gonna be wild," she said, a forecast that applies to both sides of the contest between those securing protocols and those trying to break them.

3. What an audit covers, and what it leaves behind

Audits have been the default security step in crypto for years. A project hires an outside firm to read its smart contracts line by line before deployment, the firm reports the flaws it finds, the developers fix them, and the resulting report is published as evidence of diligence. Many teams repeat the exercise with several firms or return for another round after every significant change to the code.

Monahan put numbers on how far that process goes. In her estimate, an audit deals with about 90% of the risk a project faces. Formal verification, a technique that produces a mathematical proof that the code behaves the way its designers intended, accounts for roughly another 5%. That leaves a final 5% that neither method reaches, and she warned that this remainder is where a project "might just completely collapse."

The distinction matters because of how losses in crypto tend to occur. Smart contracts frequently hold very large sums in a single place, and transactions on a blockchain are generally final once confirmed. A flaw that sits in the unreviewed fraction of the risk does not have to be common to be fatal; it only has to be found once by someone prepared to use it. Spending more on additional audits adds assurance to the portion already well covered, in Monahan's framing, while leaving the residual exposure largely where it was.

Kain Warwick, who also took part in the episode, summarized the same point in blunter terms. "Audits catch a lot of stuff," he said, but they "never catch the thing that kills you."

4. How do circuit breakers work in crypto?

A circuit breaker, as Monahan describes it, is a rule that defines an outcome that must never happen, combined with a mechanism that halts the system automatically the moment that outcome begins to occur. Instead of trying to anticipate every possible route an attacker could take, the team specifies the result it cannot tolerate and lets software enforce the limit.

She gave a sample rule to illustrate the idea: "Under no circumstances should $200 million ever leave this contract." If a transaction or a series of transactions would breach that condition, the system should "pull the plug", pausing the affected contracts before the funds are gone. The approach borrows its name and its logic from electrical panels, where a breaker trips when current exceeds a safe level, and from traditional financial markets, where trading halts are triggered when prices move too far too quickly.

The value of such a mechanism lies in the fact that it does not depend on knowing the vulnerability in advance. An unknown bug, a compromised key or a manipulated price feed can all produce the same visible symptom, which is money leaving a contract in a pattern that should not be possible in normal operation. Monahan said that when an attack is detected and answered autonomously, the team is still "gonna be fine." She described this as a layer of defense that crypto has historically lacked.

The discussion also addressed what happens when prevention is incomplete. A guest on the episode noted that a circuit breaker had helped another protocol recover funds. Monahan said the point of the mechanism is to cap the damage: a team can be "wrecked 5% of the way" rather than losing everything. In a sector where a single exploit has often emptied a protocol entirely, the difference between a partial loss and a total one can decide whether a project survives.

5. Two incidents in the same week show both outcomes

Events in the days around the episode illustrated the stakes. On Wednesday, November 18, K3 Capital paused the contracts behind sBOLD, a yield vault built on the Liquity V2 lending protocol. Security firm Blockaid said that attackers had drained about $3.43 million as of its first alert on the incident.

Two days earlier, on Monday, November 16, the perpetual futures exchange Lighter froze withdrawals after funds were drained through one trader's leaked API key. An API key is a credential that lets software place orders or move funds on a user's behalf, so a key that falls into the wrong hands can be used to act as that user. Lighter later said that its Guardian contract, a built-in safeguard, had caught the anomaly. The exchange's Security Council returned about $5 million in USD Coin (USDC), a stablecoin pegged to the dollar, to the affected trader, and withdrawals have since reopened.

The two cases are not identical, and Monahan's remarks were not presented as a verdict on either one. Taken together, though, they show the range of outcomes she was describing. In one, a pause came after millions of dollars had already left. In the other, an automated safeguard flagged unusual activity and the money was ultimately restored to its owner.

6. How North Korea-linked hackers are using AI

Monahan also addressed how the most prolific attackers in the industry are applying artificial intelligence, and her assessment was that their use of it remains limited. Hackers linked to North Korea, she said, are employing AI for minor tasks, such as adjusting the scripts they use in social engineering, the practice of deceiving employees or users into granting access.

"What they're doing with it is so boring," she said, adding that "we're not seeing the attackers optimize the AI yet." In other words, the technology is being used to polish existing methods rather than to invent new ones.

She does not expect that to last. In her view, the picture will change once younger hackers begin designing campaigns around AI from the start instead of adding it to established routines. That expectation is part of the case for building monitoring now: defenses put in place while attackers are still using the tools conservatively will already be running when their methods become more sophisticated.

Her caution about automated systems is not new. In September, Monahan said that OpenAI still could not monitor its own agents, and pointed to a breach of an Australian portal as an example.

7. What the advice means for crypto teams

For projects weighing where to spend a limited security budget, the practical message is a reordering of priorities rather than the abandonment of any one tool. Audits and formal verification remain, in Monahan's own estimate, the measures that deal with the great majority of risk. What she is urging teams to add is the capacity to observe their systems continuously and to stop them automatically when a predefined limit is crossed.

That means deciding ahead of time which outcomes are unacceptable, expressing them as rules that software can check, and granting an automated system the authority to pause contracts without waiting for a person to wake up, read an alert and gather signatures. It also means accepting that a pause triggered in error is a smaller cost than a drain that runs to completion.

Monahan's argument is that the tools to do this are no longer reserved for large organizations. With AI assisting in the work of watching for anomalies, she said, small teams can build reactive systems of their own, and with AI agents probing projects more frequently, she expects the months ahead to test which teams have done so.

Related Articles

NEWSLETTERS

Don't miss another story.

Subscribe to the MINRK Newsletter today.

Tick the newsletters you want. No selection? You'll get MINRK Headlines.

By signing up, you will receive emails about MINRK products and you agree to our terms of use and privacy policy.

MINRK Daybook

Market analysis for crypto traders and investors.

EVERY WEEKDAY

MINRK for Advisors

Defining crypto, digital assets and the future of finance for financial advisors.

EVERY THURSDAY

MINRK Protocol

Exploring the tech behind crypto one block at a time.

WEEKLY

MINRK Long & Short

A must read for institutions. Insights, news and analysis delivered weekly.

EVERY WEDNESDAY

MINRK Headlines

The biggest crypto news and ideas of the day.

EVERY WEEKDAY

MINRK Policy

Examining the intersection of cryptocurrency and government.

WEEKLY

MINRK Research

Join thousands of readers who rely on MINRK for data-driven insights on the latest digital asset trends.

MONTHLY