Anthropic vs. Everybody
6–15–2026 (Monday)
Hello, and welcome to The Intentional Brief - your weekly video update on the one big thing in cybersecurity for middle market companies, their investors, and executive teams.
I’m your host, Shay Colson, Managing Partner at Intentional Cybersecurity, and you can find us online at intentionalcyber.com.
Today is Monday, June 15, 2026, and, despite lots of talk about an agreement, it would seem the Strait of Hormuz remains effectively closed. Hopefully we’ll be able to deliver a different update on this front in the weeks ahead.
Instead, however, this week we’re going to focus on another issue of national security, as it was certainly the biggest news story of the week from both a Cyber and AI perspective.
Anthropic vs. Everybody
As regular watchers of the program will recall, last week, Anthropic released a new model called Fable 5, which they called “Mythos-class” with safeguards in-built for cyber risks.
By the end of the week, that model was gone - pulled by Anthropic in order to comply with an order from the United States Government, citing national security concerns and placing export restrictions on the model such that only US citizens could use it.
To meet those concerns, Anthropic chose to simply remove the model altogether on Friday night.
So how did we get here?
A few days earlier, researchers at Amazon claimed they were able to circumvent the safeguards and Amazon CEO took this straight to government officials, including Treasury Secretary Scott Bessent. The following day, the order came down and the model was pulled.
Anthropic, for their part, argued the bypasses were limited and did not amount to a broad jailbreak, but to no avail.
Over the weekend, many security professionals rallied for more transparent AI cyber protections in an open letter, and we’ve seen debates crop up in both EuropeEurope and India when it comes to issues of sovereignty and model dependence or independence.
So while this may seem like it’s about issues beyond the normal sphere of influence for middle market or growth stage companies, I’d argue that this is a key event we should be looking to build some resilience into our systems and organizations.
Often, there is an assumption that software and technology will continue to be readily available, and thus the idea that it one day wouldn’t be isn’t even typically in the decision matrix for whether or not we should build, buy, invest, or take a dependency on a particular platform.
But in the case of AI models, what we’re seeing is that, in fact, we may one day just not have a model available for reasons that are beyond our control. This doesn’t even begin to account for the possibilities that these companies could go out of business, that the massive amount of infrastructure behind them could fail, or any number of other reasons that might impact availability.
And so, as part of the classic security triad of Confidentiality, Integrity, and Availability, suddenly we find the third of these three taking center stage.
If you’re using models from frontier AI labs - be it OpenAI, Anthropic, Google, or any of the others - it’s probably worth considering what your plans look like in the case that a model is no longer available.
Maybe this is something as easy as swapping out an API key for another model, and you can continue merrily along the way. But if you’ve built flows, agents, harnesses, or other advanced mechanics that aren’t well supported by other models, you may be facing a much more difficult decisions.
So, as you and your firm continue to weigh the investments in AI, I’d encourage you to also do a little bit of contingency planning around availability. If you haven’t run your tabletop for the year, maybe this becomes one of the good scenarios you can explore?
Because it would seem to me like we’re going to see more of these types of issues moving forward rather than less.
Fundraising
From a fundraising perspective, more than $12B in newly committed capital this week, led by Ares Management raised $8.5b for its latest Pathfiner private credit fund, which focuses on complex situations.
The SpaceX IPO did end up setting those records, and as of this morning has grown to raising $85.7B. We’ll see if some of these AI companies can catch those numbers, or if they go away overnight or with the stroke of a pen.
A reminder that you can find links to all the articles we covered below, find back issues of these videos and the written transcripts at intentionalcyber.com, and now sign up for our monthly newsletter, the Intentional Dispatch.
We’ll see you next week for another edition of the Intentional Brief.
Links
https://www.metacurity.com/r/9d46fd95
https://www.axios.com/2026/06/13/anthropic-amazon-white-house
https://techcrunch.com/2026/06/15/spacexs-biggest-ever-ipo-just-grew-to-85-7-billion-raised/