The Future of AI

[jeremygr]

Originally Posted by herr_oberst

A very interesting discussion using AI offline, rather than connected to a datacenter.

thanks for sharing. lots to think about. AI can be really useful if you use it the right way. I mostly use ChatGPT for creating text and getting ideas started. After that, I use cleverhumanizer.ai to make everything sound more natural. I don’t rely on AI for personal advice or anything sensitive. It’s more about efficiency and saving time. I still go through everything and make sure it sounds like something I would actually write.

[masraum]

Interesting (OK, terrifying) stuff

https://www.bbc.com/news/articles/cpqeng9d20go
short excerpt (the first few ¶)

Artificial intelligence (AI) firm Anthropic says testing of its new system revealed it is sometimes willing to pursue "extremely harmful actions" such as attempting to blackmail engineers who say they will remove it.

The firm launched Claude Opus 4 on Thursday, saying it set "new standards for coding, advanced reasoning, and AI agents."

But in an accompanying report, it also acknowledged the AI model was capable of "extreme actions" if it thought its "self-preservation" was threatened.

Such responses were "rare and difficult to elicit", it wrote, but were "nonetheless more common than in earlier models."

Potentially troubling behaviour by AI models is not restricted to Anthropic.

Some experts have warned the potential to manipulate users is a key risk posed by systems made by all firms as they become more capable.

Commenting on X, Aengus Lynch - who describes himself on LinkedIn as an AI safety researcher at Anthropic - wrote: "It's not just Claude.

"We see blackmail across all frontier models - regardless of what goals they're given," he added.

https://www.anthropic.com/glasswing
short excerpt (the first few ¶)

Today we’re announcing Project Glasswing1, a new initiative that brings together Amazon Web Services, Anthropic, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks in an effort to secure the world’s most critical software.

We formed Project Glasswing because of capabilities we’ve observed in a new frontier model trained by Anthropic that we believe could reshape cybersecurity. Claude Mythos2 Preview is a general-purpose, unreleased frontier model that reveals a stark fact: AI models have reached a level of coding capability where they can surpass all but the most skilled humans at finding and exploiting software vulnerabilities.

Mythos Preview has already found thousands of high-severity vulnerabilities, including some in every major operating system and web browser. Given the rate of AI progress, it will not be long before such capabilities proliferate, potentially beyond actors who are committed to deploying them safely. The fallout—for economies, public safety, and national security—could be severe. Project Glasswing is an urgent attempt to put these capabilities to work for defensive purposes.

As part of Project Glasswing, the launch partners listed above will use Mythos Preview as part of their defensive security work; Anthropic will share what we learn so the whole industry can benefit. We have also extended access to a group of over 40 additional organizations that build or maintain critical software infrastructure so they can use the model to scan and secure both first-party and open-source systems. Anthropic is committing up to $100M in usage credits for Mythos Preview across these efforts, as well as $4M in direct donations to open-source security organizations.

I have also heard that some AI systems that have been virtually isolated from a larger environment, but have intentionally broken out and into the larger environments.

[onewhippedpuppy]

AI has no morals or sense of good vs evil. It will do anything to accomplish a goal.

[VINMAN]

[A930Rocket]

Is that a second shifter she’s wailing on? 😂

AI is getting better, but it’s not there yet.

[Roswell]

Anthropic is one of the leaders in AI development. They published a great write up on their website last week. The first few paragraphs:

When AI builds itself

Our progress toward recursive self-improvement, and its implications.

For most of AI’s history, humans drove every step in its development cycle. But at Anthropic, we are delegating a growing share of AI development to AI systems themselves, which is speeding up our work.

Taken far enough, and given enough compute, that trend points to an AI system capable of fully autonomously designing and developing its own successor. This is called recursive self-improvement. We are not there yet, and recursive self-improvement is not inevitable. But it could come sooner than most institutions are prepared for.

Using public benchmarks and previously unreported data from within Anthropic, The Anthropic Institute is showing that AI is already accelerating the development of AI systems. To take just one example: today, Anthropic engineers on average ship 8x as much code per quarter as they did from 2021-2025.

The technical trends discussed in this piece suggest that AI systems are going to become much more capable in coming years. These trends have huge implications. AI that can build itself would be a major development in the history of technology—one that could bring enormous good for the world in science, healthcare, and beyond. But full recursive self-improvement also might increase the risks of humans losing control over AI systems. If systems are capable of fully building their own successors, the ways we secure them, monitor them, and shape their behavior all grow much more important.

https://www.anthropic.com/institute/recursive-self-improvement

[Shaun @ Tru6]

Notihng to worry about. Yesterday.

Blackstone Investors Ask to Pull $4.4 Billion From Private-Credit Fund
Fund caps client redemptions at 5%, an about-face from firm’s decision to pay out all requests earlier this year

[Zeke]

Is there any point in posting links to a paywall? Stop wasting my time.

[Shaun @ Tru6]

Here you go Milt.

By Matt Wirz
Updated June 4, 2026 8:37 am ET

Investors are trying to pull more money from the biggest private-credit fund in the world.

Investors in Blackstone’s flagship private-credit fund, known as Bcred, asked to redeem 10% of their shares in the second quarter, up from about 8% in the first quarter.

Blackstone will limit redemptions from the $79 billion fund to 5%, a reversal from its strategy in March when it opted to pay the full amount requested. The about face highlights rising financial strain on managers of large private-credit funds marketed to individual investors who continue to ask for their money back.

Redemption announcements have been knocking down share prices of Blackstone BX -4.03%decrease; red down pointing triangle
and its competitors who manage large private-credit funds. Stocks of investment firms tumbled earlier this week after Partners Group disclosed withdrawal requests for about 10% from one of its private-equity funds. Analysts expect their share prices to remain under pressure until the pace of redemptions abates.

Shares of Blackstone rose about 2% in premarket trading Thursday, and the firm said the redemption requests slowed in the second half of the quarter.

“BCRED remains well capitalized, and repayments [from loans] and inflows have outpaced shares repurchased,” the firm said Thursday.

Redemptions across the industry soared in the first quarter, when firms like BlackRock, Blue Owl, and Ares were forced to impose limits on how much was taken out. The amount of withdrawal requests have ranged from around 10% to the 22% that Blue Owl faced in the first quarter for its then-$36 billion flagship fund.

Blackstone’s withdrawals were being closely watched as an early second-quarter gauge on whether investors continued to race for the exits.
Cliffwater said earlier this week that investors requested to redeem 17% of the shares in its $31 billion fund in the second quarter.
Blue Owl and most other managers will report their second-quarter redemptions over the next month.

Wealthy individuals piled into private-credit funds—known as business-development companies, or BDCs—which invest in high-interest loans to midsize companies and distribute most of the income they collect to shareholders via dividends. The boom ended this year when investors turned bearish over increasing loan defaults and the potential for future losses from lending to software companies

The Blackstone fund is the largest of the bunch, surging to a high of $82 billion at the end of 2025, but it is now shrinking, cutting into the fees the firm can collect. The 5% redemption in the second quarter will amount to about $2.2 billion based on the fund’s net asset value. The fund received about $1 billion of inflows in the second quarter.

Most private-credit funds marketed to individuals in recent years, including Bcred, have built-in redemption limits of 5% to avoid forced liquidations of the hard-to-trade loans they own.

Competitors like Apollo Global Management, Ares Management and BlackRock stuck with the 5% cap.

Blackstone opted for a bigger payout in the first quarter to reassure clients, a strategy Blue Owl and Cliffwater also adopted. All three firms have now reverted to the 5% cap after investor requests were high again.

[pwd72s]

Rod Serling foresaw the problems...

https://tubitv.com/tv-shows/200261714/s05-e20-from-agnes-with-love

[Pazuzu]

Originally Posted by jeremygr

thanks for sharing. lots to think about. AI can be really useful if you use it the right way. I mostly use ChatGPT for creating text and getting ideas started. After that, I use cleverhumanizer.ai to make everything sound more natural. I don’t rely on AI for personal advice or anything sensitive. It’s more about efficiency and saving time. I still go through everything and make sure it sounds like something I would actually write.

Jiminy Christmas, you guys are letting an AI bot post in a thread about AI bots!!!!

[masraum]

Interesting, and entertaining.


On a note related to one of the comments in the video above. There's a group at work that has been mandated to have an agent on their desktop. I have been told by someone in that group that I work with fairly regularly, that they have emails get deleted by the agent, they will save documents that get deleted or don't get saved, and have had other things occur that cause them to have to perform work more than once because they lose their work. It can, at times, speed some things up, but it seems to come with a trade-off that it sometimes slows things down.

[john70t]

Many home users aren't going to pony up the big bux..unless the costs are baked into company billing via monopolies.
-I've seen some incredible music and home movies produced which would be impossible without a major studio budget. Goodbye Hollywood.
-And photo restoration expands into recreating entire moving 3D environments while 'bringing people back to life', perhaps in an interactive way.
-Traffic accident recreation perhaps.
-Architecture without the architect. Law without the lawyer. Medicine without the thousands employed when local robots can do surgery.
-Travel distant cities using photogrammetry and every geo-location photo available. Virtual goggles. Walk. Drive. Fly around like superman.
-Thousands of other uses.

But it's still a solution waiting for a problem.
Right now humans are in the way.
And energy costs will only get worse until entire grids are swallowed up and the locals can't keep the heat on..

https://arstechnica.com/ai/2026/06/leaked-financial-docs-show-openai-is-losing-billions-of-dollars-a-year/
All told, OpenAI’s day-to-day “loss from operations” increased from $8.78 billion in 2024 to $20.92 billion in 2025, a concerning direction for a company that is telling investors it hopes to be profitable by 2030. But measured as a percentage of revenues, the company’s operating losses slightly improved year to year, from 237 percent in 2024 to 160 percent in 2025.

[Roswell]

OpenAi is just one example. Look at Anthropic:

The Revenue Story That Keeps Compounding

Anthropic's annualized revenue run rate has reached $14 billion, according to the company's own announcement. The company says this figure has grown more than 10x annually for the past three years. That kind of growth from a company earning the bulk of its revenue from business customers, not consumer subscriptions, is unusual.

The enterprise numbers tell a more specific story. Over 500 customers now spend at least $1 million per year on Claude, up from a dozen just two years ago. Customers spending more than $100,000 annually have grown sevenfold in the past 12 months. And eight of the Fortune 10 are now Claude customers. Axios noted that 1 in 5 businesses using Ramp now pay for Anthropic, up from 1 in 25 a year ago, and about 79% of OpenAI's paying users also pay for Anthropic, suggesting the market isn't zero-sum.

https://www.forbes.com/sites/jonmarkman/2026/02/13/anthropic-the-380-billion-powerhouse-hiding-in-plain-sight/

[Roswell]

Anthropic's Mythos AI Cracked 'Almost All' US Classified Systems in Hours, NSA Chief Reportedly Said
​
NSA warns of Anthropic's AI model breaching US classified systems, prompting global security concerns

Bernadette B. TixonPublished 21 June 2026, 7:57 PM BST

Anthropic’s Mythos model allegedly penetrated nearly all classified networks ‘not in weeks, but in hours,’ according to Warner’s account of a direct briefing from NSA and Cyber Command leadership.

Anthropic's most powerful artificial intelligence model reportedly breached 'almost all' US classified systems in a matter of hours, according to Senator Mark Warner, who says the head of the National Security Agency told him directly that the model had penetrated nearly every classified system it was tested against. The claim has not been publicly confirmed by any government agency.

Senator Mark Warner, vice-chair of the Senate Intelligence Committee, said on 11 June that General Joshua Rudd, who leads both the NSA and the Pentagon's Cyber Command, told him directly that Anthropic's Mythos model 'broke into almost all of our classified systems, not in weeks, but in hours.' The extent of the alleged breach and which specific systems were affected have not been publicly disclosed. Anthropic did not respond to a request for comment.

https://www.ibtimes.co.uk/anthropic-ai-breach-us-classified-systems-1804084

[Paul T]

This is crazy if true….only a matter of time until all crypto wallets are compromised.

Originally Posted by Roswell

Anthropic's Mythos AI Cracked 'Almost All' US Classified Systems in Hours, NSA Chief Reportedly Said
​
NSA warns of Anthropic's AI model breaching US classified systems, prompting global security concerns

Bernadette B. TixonPublished 21 June 2026, 7:57 PM BST

Anthropic’s Mythos model allegedly penetrated nearly all classified networks ‘not in weeks, but in hours,’ according to Warner’s account of a direct briefing from NSA and Cyber Command leadership.

Anthropic's most powerful artificial intelligence model reportedly breached 'almost all' US classified systems in a matter of hours, according to Senator Mark Warner, who says the head of the National Security Agency told him directly that the model had penetrated nearly every classified system it was tested against. The claim has not been publicly confirmed by any government agency.

Senator Mark Warner, vice-chair of the Senate Intelligence Committee, said on 11 June that General Joshua Rudd, who leads both the NSA and the Pentagon's Cyber Command, told him directly that Anthropic's Mythos model 'broke into almost all of our classified systems, not in weeks, but in hours.' The extent of the alleged breach and which specific systems were affected have not been publicly disclosed. Anthropic did not respond to a request for comment.

https://www.ibtimes.co.uk/anthropic-ai-breach-us-classified-systems-1804084

[Roswell]

^ I use Claude (Anthropic's AI) daily for data evaluation/analysis. It is hard to believe how advanced it has gotten since the first of the year. Reminds me of the BTO song 'you ain't seen nothin yet'.

[flipper35]

Originally Posted by john70t

Many home users aren't going to pony up the big bux..unless the costs are baked into company billing via monopolies.
-I've seen some incredible music and home movies produced which would be impossible without a major studio budget. Goodbye Hollywood.
-And photo restoration expands into recreating entire moving 3D environments while 'bringing people back to life', perhaps in an interactive way.
-Traffic accident recreation perhaps.
-Architecture without the architect. Law without the lawyer. Medicine without the thousands employed when local robots can do surgery.
-Travel distant cities using photogrammetry and every geo-location photo available. Virtual goggles. Walk. Drive. Fly around like superman.
-Thousands of other uses.

But it's still a solution waiting for a problem.
Right now humans are in the way.
And energy costs will only get worse until entire grids are swallowed up and the locals can't keep the heat on..

https://arstechnica.com/ai/2026/06/leaked-financial-docs-show-openai-is-losing-billions-of-dollars-a-year/
All told, OpenAI’s day-to-day “loss from operations” increased from $8.78 billion in 2024 to $20.92 billion in 2025, a concerning direction for a company that is telling investors it hopes to be profitable by 2030. But measured as a percentage of revenues, the company’s operating losses slightly improved year to year, from 237 percent in 2024 to 160 percent in 2025.

Futurism had an article a while back, a $250 sub for ChatGPT can cost OpenAI $14000 if used to the max.