JPMorgan Security Chief: Weigh Taking Services Off Internet

Pat Opet, whose X profile reads “security @jpmorgan @chase”, wrote on November 7 that people working in cybersecurity “should be evaluating your options to abstract your companies services from the internet” (post on X).

The post’s reasoning: “why allow an unauthenticated anonymous agent of an attacker to interact with your systems?” The post had 100,672 views by 01:35 UTC on November 8, about eight hours after it went up.

Business Pill · LEAST PRIVILEGE

A short Business Engineer explainer on least privilege: giving an AI agent only the access its task needs. It teaches the general idea only and says nothing about JPMorgan or this post.

What the JPMorgan Security Chief Wrote

The post is three lines long. After the question about attacker agents, it ends: “consider your architectural levers now, you’re going to need them”. It does not name a product, a vendor, a timeline or a specific JPMorgan system, and it does not say what JPMorgan itself is changing.

In plain terms, abstracting a service from the internet means it no longer answers anyone who finds its address; a user or system has to be identified before it can reach the service at all. The post does not say how JPMorgan would do that.

What Was Posted on X

Pat Opet, whose X profile reads ‘security @jpmorgan @chase’, posted on November 7 that people working in cybersecurity should be evaluating options to abstract company services from the internet, asking why an unauthenticated anonymous agent of an attacker should be able to interact with their systems. This piece is built on that post.

Ed Sim quoted Opet’s post the same evening, introducing Opet as JPMorgan Chase CISO and summarising the idea as ‘private by default’.

How It Was Read

Ed Sim quoted the post the same evening, introducing Opet as “JPMorgan Chase CISO”, and summarised it as “private by default” (post on X). “we spent decades making everything reachable and building security around it”, Sim’s post said, and “now, with autonomous AI agents, we may spend the next decade making only what needs to be reachable visible at all”. Sim’s post had 32,991 views by the same time.

Why Banks Are Watching Agents

On October 22, Axios reported that major US banks were detecting unauthorized third-party AI agents inside customer accounts, and that JPMorgan, Bank of America and Citi were restricting them while Morgan Stanley had begun blocking them, as we reported. Those agents usually log in with the customer’s own credentials.

On November 6, Microsoft said a financially motivated group it tracks as Storm-3358 had used the ARTEX agentic AI framework since October against financial organizations, scanning systems connected to the public internet (our piece). On November 5 it described one case in which an AI agent went from scanning GitHub for secrets to taking data from Azure Key Vault in five minutes (our piece).

Our Numbers

By our arithmetic, 16 days separate the Axios report that JPMorgan was restricting outside agents in customer accounts (October 22) and its security chief’s post about attacker agents (November 7). Opet’s post had about three times the views of Ed Sim’s quote of it, 100,672 against 32,991, by our arithmetic.

The Structural Read

As we read the post, it is about who can reach a system at all, not about how to defend a system everyone can reach.

As we read it alongside Axios’s October 22 report, JPMorgan’s restrictions on outside AI agents in customer accounts and its security chief’s post both concern agents reaching bank systems, our reading.

As we read the post, it gives no JPMorgan plan, product or date; it is a personal post on X addressed to people working in cybersecurity.

Three Implications

FOR BANKS Axios reported on October 22 that JPMorgan, Bank of America and Citi were restricting outside AI agents in customer accounts, and Morgan Stanley had begun blocking them.

FOR SECURITY TEAMS Opet’s post tells people working in cybersecurity to consider their architectural levers now, and says they are going to need them.

FOR AGENT BUILDERS Microsoft said on November 6 that a financially motivated group used the ARTEX agentic AI framework to scan systems connected to the public internet at financial organizations.

The Business Engineer Lens

This story maps onto the Business Engineer framework The Agentic Harness War.

The analysis says: “The agentic surface — the harness — is becoming where value pools in the entire AI stack.”

As we read it, Opet’s post looks at agents from the other side: whether an unauthenticated agent should reach a company’s systems at all. The analysis discusses AI products and does not discuss JPMorgan, Opet or this post.

What Is Not Established

The post is a personal post on X, not a JPMorgan announcement, and it does not say whether JPMorgan has moved or will move any service off the public internet. We found no JPMorgan statement on it. This piece makes no forecast and contains no security advice.

The Bottom Line

JPMorgan’s security chief told the cybersecurity industry on November 7 to evaluate taking company services off the public internet, asking why an attacker’s anonymous agent should be able to reach them. It follows Axios’s October report on banks restricting outside agents and Microsoft’s reports of an AI attack framework aimed at financial firms.