Catholic, All things Identity Management, Stealth Founder, Selling products to Fortune 500 in pajama shorts. Building products via arguing with AI tools.
Saylor’s entire equity position is in MSTR common. Which gives him a massive incentive to sell bitcoin to protect the prefs. But more importantly it means he will sell BTC before seeing his personal wealth collapse. But the real angle is the convexity of bankruptcy. The underlying asset is not impaired simply because the owner can’t operate. A drawn out restructuring ironically helps the underlying asset because it can’t be forced into liquidation which iconically reduces the float. With an asset with a beta of 1.5, there is a non trivial likelihood that the underlying asset appreciators over the course of restructuring, especially considering the BTC crash you probably when MSTR enters restructuring. My view is it’s a convexity arbitrage.
I have been in identity management for 15 years and recently took on a project working with the SOC and Pentest team. I have to tell you zero trust scares the shit out of me when combined with “identity is the perimeter” the industry is trying to sell. Thats exactly opposite of defense in depth. My old mentors in the Defense industry would never have allowed such a dangerous strategy.
@ZackKorman I find if I start every prompt with “bro” it is more empathetic, no joke. Throw in a “Bro I’m a vibe coder help me” and it’s empathy super mode, you can see it in the chain of thought.
Why does IT security care more about the ai tools their users are using than they do about being able to identify an attack from an AI enabled attacker. All you are going to see is shell commands. Also how many vulnerabilities are we creating with observability tools. Copilot keeps a log of every chat your users ever had. If someone gets access to those logs your done.
Over the last 3 years every boomer manager in the country has demanded a return to office in order to boost creativity and productivity. The average firm spends $10,000 a year per employee on office leasing. These same boomers choke on why we would want to spend $1000 a month on AI models. Any company could be a leader in AI overnight if they went 100% remote. WFH is going to be the only way to survive AI costs in the coming years.
@PeteHegseth@DeptofWar@AnthropicAI Did GPT 5.5 develop the plan on how to keep the Strait of Hormuz open? You probably want to use extra high think on the next invasion. I find it’s also helpful to do an adversarial review with more than one sub agent.
36 hours until the mother of all bubbles bursts if the Feds don’t turn fable 5 back on. We don’t need the music to stop, if AI demand just slows this market is done. I can’t think of any cyber incident that would be more destructive than that short of complete shut down of the grid. The precedent this sets is catastrophic to risk capital. This exponential changes the terminal value of every dollar invested in Ai, ai infrastructure and every tangential industry. If these IPO windows get pushed back even a month or two these companies have to adjust their token subsidies, which curbs demand and creates a death spiral. If the cyber defenders can’t do their job then that needs to be exposed now. We need to surface these weaknesses now before the open source models are doing the except same thing within 90 days.
Yes absolutely. The impact a stock market crash is far more damaging to the average American than moderate cyber attack. I think “global safety” also has to be quantified. If there is evidence of imminent threat that creates a clear and present danger or significant loss of life that is a fair argument. None of that evidence has been presented. If that is the case they should have shut down Fable immediately. Reports are these discussions have been occurring since at least Thursday. The Feds waited until the futures markets closed at 5pm eastern to make this announcement. If there was an ongoing incident they were aware of I expected they would have moved much faster and not worried to wait for markets to close. This smells like boomer alarmism about a hypothetical. But I would also argue that the only way to identify many of these cyber vulnerabilities is through gradual exposure. The open source models are only months behind. I have worked in cyber for 17 years. Today I can get any of the leading models to drive Kali Linux and perform any numbers of attacks without any need to jailbreak them. The models happily do it. These attacks are coming and they are possible today with open source models.