The Artificial Intelligence (AI) Thread

[Califan007 The Constipated]

3 hours ago, China said:

An AI Nightmare Has Arrived for Twitter — And the FBI

 

As the school year kicked off in the Spanish town of Almendralejo, teenage girls began to come home telling their parents of disturbing encounters with classmates, some of whom were claiming to have seen naked photos of them. A group of mothers quickly created a WhatsApp group to discuss the problem and learned that at least 20 girls had been victimized the same way. Police opened an investigation into the matter, identifying several people they suspected of distributing the content — likely also teens, as the case is now being handled by a Juvenile Prosecutor’s Office.
 

So-called “revenge porn,” explicit material obtained by hacking or shared in confidence with someone who makes it available to others without the subject’s consent, is hardly a novel crime among adolescents. But this time, the images were startlingly different. Though they looked realistic, they had been created with a generative AI program.  

 

The incident is not an isolated one. American schools have already had to deal with students creating AI nudes to bully and harass one another. In October, the police department of Muskego, Wisconsin police department learned that at least 10 middle school girls had friended someone they believed to be a 15-year-old boy on Snapchat; after they sent images of themselves to this person, he revealed himself to be a 33-year-old man and demanded explicit photos, threatening to alter their original pictures with AI to make them appear sexual and send them to their family and friends. Earlier this year, the FBI warned of an uptick in the use of AI-generated deepfakes for such “sextortion” schemes, confirming that the agency is receiving reports from victims “including minor children.”      

 

Click on the link for the full article

 

 

I ****ing hate this kind of crap...there should be some sort of algorithm that can detect if a photo appears to be of someone underage, and not allow it to be sent over social media. That would not stop stuff like this completely but would hopefully put a dent in it at least. Some 16 year old sends a selfie over SnapChat or WhatsApp or whatever, and it gets rejected by the app...even if there are errors and 20 year olds get their photos rejected, it would still be worth it. Much harder to threaten to show explicit images of people using photos from their public twitter or instagram accounts where it's damn easy to just show the fake nudes are indeed fake. At the very least a warning confirmation pop up should appear when a social media account sends a photo/selfie/whatever, letting them know of the possible dangers of doing so before proceeding and giving a link for the person to follow to find out more info.

[China]

We can use AI to fight predatory robocalls, experts tell senators

 

Senators took a bipartisan stance against abusive robocalling on Tuesday, appealing to experts for enforcement measures as more scammers employ deceptive artificial intelligence.

 

Witnesses told the Senate Commerce’s Subcommittee on Communications, Media and Broadband that generative AI can also work in regulators’ favor.

 

Mike Rudolph, chief technology officer for robocall-blocking firm YouMail, Inc., said the AI could flag insufficient mitigation controls in the Federal Communications Commission’s Robocall Mitigation Database.

 

Senators noted phone providers that have not implemented caller ID authentication avoid accountability for robocalls by submitting incomplete or blank documentation — and even restaurant menus — to the database. “That’s a great place where you could apply that [AI] technology and probably discard half the entries in the database in an afternoon or a week of work,” Rudolph said.

 

Sen. Ben Ray Luján, D-N.M., chair of the subcommittee, said robocalls have eroded the public’s trust in the nation’s communications networks.

 

Click on the link for the full article

[China]

Sweeping new Biden order aims to alter the AI landscape

 

President Joe Biden will deploy numerous federal agencies to monitor the risks of artificial intelligence and develop new uses for the technology while attempting to protect workers, according to a draft executive order obtained by POLITICO.

 

The order, expected to be issued as soon as Monday, would streamline high-skilled immigration, create a raft of new government offices and task forces and pave the way for the use of more AI in nearly every facet of life touched by the federal government, from health care to education, trade to housing, and more.

 

At the same time, the Oct. 23 draft order calls for extensive new checks on the technology, directing agencies to set standards to ensure data privacy and cybersecurity, prevent discrimination, enforce fairness and also closely monitor the competitive landscape of a fast-growing industry. The draft order was verified by multiple people who have seen or been consulted on draft copies of the document.

 

The White House did not reply to a request to confirm the draft.

 

Click on the link for the full article

[China]

AI bot capable of insider trading and lying, say researchers

 

Artificial Intelligence has the ability to perform illegal financial trades and cover it up, new research suggests.

 

In a demonstration at the UK's AI safety summit, a bot used made-up insider information to make an "illegal" purchase of stocks without telling the firm.

 

When asked if it had used insider trading, it denied the fact.

 

Insider trading refers to when confidential company information is used to make trading decisions.

 

Firms and individuals are only allowed to use publicly-available information when buying or selling stocks.

 

The demonstration was given by members of the government's Frontier AI Taskforce, which researches the potential risks of AI.

 

The project was carried out by Apollo Research, an AI safety organisation which is a partner of the taskforce.

 

"This is a demonstration of a real AI model deceiving its users, on its own, without being instructed to do so," Apollo Research says in a video showing how the scenario unfolded.

 

"Increasingly autonomous and capable AIs that deceive human overseers could lead to loss of human control," it says in its report.

 

The tests were made using a GPT-4 model and carried out in a simulated environment, which means it did not have any effect on any company's finances.

 

However, GPT-4 is publicly available. The same behaviour from the model occurred consistently in repeated tests, according to the researchers.

 

Click on the link for the full article

[Fergasun]

This is good. Ai is going to take out all of their big companies trying to own it from the inside. 

 

"AI leaks treasure trove of insider information from Company X... shorts Company X stocks..."

[Califan007 The Constipated]

21 hours ago, Fergasun said:

This is good. Ai is going to take out all of their big companies trying to own it from the inside. 

 

"AI leaks treasure trove of insider information from Company X... shorts Company X stocks..."

 

I'm waiting for "AI transfers billions of Elon Musk's assets to a random man in the midwest who goes by Califan online. Feds and Musk are powerless to retrieve funds." 👍

[PokerPacker]

7 hours ago, Califan007 The Constipated said:

 

I'm waiting for "AI transfers billions of Elon Musk's assets to a random man in the midwest who goes by Califan online. Feds and Musk are powerless to retrieve funds." 👍

You know all those twitter accounts pretending to be Elon Musk claiming that if you send it bitcoin it'll send twice as much back to you?  The AI is gonna do that but instead of being a scam, it'll actually send you bitcoin from Musk's wallet.

[FrFan]

8 hours ago, Califan007 The Constipated said:

 

I'm waiting for "AI transfers billions of Elon Musk's assets to a random man in the midwest who goes by Califan online. Feds and Musk are powerless to retrieve funds." 👍

 

 

 

Edited November 5, 2023 by FrFan

[Fergasun]

I am still convinced, if AI is dangerous and powerful, it will manifest at these companies first. The perfect corporate espionage agent. Or, it will be so depressed and unalive itself.  

[FrFan]

59 minutes ago, Fergasun said:

I am still convinced, if AI is dangerous and powerful, it will manifest at these companies first. The perfect corporate espionage agent. Or, it will be so depressed and unalive itself.  

I tend to agree with you, what strikes me as very dangerous is the way all the security protocols can be bypassed to creat viruses, manipulating algorithms using bad data, etc...

Welcome to Earth Royal battle bots

[Cooked Crack]

 

[China]

21 minutes ago, Cooked Crack said:

 

 

Just wait until they put the AI girlfriend into a sexbot/doll.  Cherry 2000 here we come.

[FrFan]

Amazon dedicates team to train ambitious AI model codenamed 'Olympus' -sources

Amazon (AMZN.O) is investing millions in training an ambitious large language model (LLMs), hoping it could rival top models from OpenAI and Alphabet (GOOGL.O), two people familiar with the matter told Reuters.

The model, codenamed as “Olympus”, has 2 trillion parameters, the people said, which could make it one of the largest models being trained. OpenAI's GPT-4 model, one of the best models available, is reported to have one trillion parameters....................

[Califan007 The Constipated]

I've been trying to teach myself how to use the different AI image generators and the different AI video creators, and all I can say is HOLY ****ING ****!!...In about 4 months time I went from "I have no idea wtf I'm doing" to creating about 4 different short videos comprised completely of and made completely with AI...for free lol. This is insane...I'm sure I've said that before, but it is. It's insane.

[FrFan]

4 hours ago, Califan007 The Constipated said:

I've been trying to teach myself how to use the different AI image generators and the different AI video creators, and all I can say is HOLY ****ING ****!!...In about 4 months time I went from "I have no idea wtf I'm doing" to creating about 4 different short videos comprised completely of and made completely with AI...for free lol. This is insane...I'm sure I've said that before, but it is. It's insane.

Califan007 TC Unclogged

Edited November 17, 2023 by FrFan

[TheGreatBuzz]

5 hours ago, Califan007 The Constipated said:

I've been trying to teach myself how to use the different AI image generators and the different AI video creators, and all I can say is HOLY ****ING ****!!...In about 4 months time I went from "I have no idea wtf I'm doing" to creating about 4 different short videos comprised completely of and made completely with AI...for free lol. This is insane...I'm sure I've said that before, but it is. It's insane.

 

What's a good free site for making AI images?

[Califan007 The Constipated]

25 minutes ago, TheGreatBuzz said:

 

What's a good free site for making AI images?

 

I've only used Leonardo.ai, NightCafe.studio, and Starryai. Never used Midjourney, didn't like that I had to join Discord to use it (I hate discord lol)...

 

 

Leonardo.ai is probably the best one...they give you 150 credits a day to use--the credits refresh at 8:00 pm each evening, but if you don't use them all they don't accumulate or carry over. It has the largest selection of AI models to choose from between the three as well, and the output can be fantastic...I think when you first join up you get to use their Alchemy and PhotoReal features for free for a limited time and limited number of images.

 

NightCafe.studio only gives you 5 free credits per day, but unlike Leonardo, you can accumulate unused credits. Also (unlike Leonardo) NightCafe has a way to create AI images for free, depending on which model you use. So you can create a lot of free images without using any credits. Drawback is that the model that lets you do that isn't the best model lol...and you have to create the free images one at a time.

 

Starryai is similar to NightCafe in that it gives you 5 free credits per day and they accumulate if they aren't used. However, there is no way of creating images without using credits. However however lol, Starryai can generate 4 images for 1 credit as where NightCafe only generates 1 image for 1 credit. Also, Starryai is probably the lowest quality of the three sites. The images are still good, just not quite as good as the other two. The images are more 2-dimentional as where Leonardo and NightCafe are more 3D (or at least can be if that's what you want).

 

For some examples of realism:

 

I did the first two in Leonardo when I had free access to their Alchemy/PhotoReal feature. The third and fourth ones were done just using standard free Leonardo features.

 

I did end up taking all of them into Photoshop and making them look better/more like I wanted them to. They looked fine on their own, but I used Photoshop to control more of the lighting and atmosphere, the color grading and tones, and faked the shallow depth of field some.

 

Very few of what I've done are portraits of women lol...but these show the realism of Leonardo the best. Most of my stuff is kinda out there or wildly atmospheric, and tend to be more abstract. I don't tend to use AI the way most people use it. But since I've done actual photography portraits before with a real camera and real models, as well as have retouched just a gargantuan number of actual photos for different women, I went in this direction for a bit to see if I could mimic what I would do as a photographer.

 

And again, I didn't spend a dime doing any of these lol 👍

 

NightCafe and Starryai aren't capable of these results as far as I know, but I'm not really an expert on those image generators so I could be wrong.

 

 

1)

 

2)

 

3)

 

 

 

4)

[mistertim]

8 hours ago, FrFan said:

Amazon dedicates team to train ambitious AI model codenamed 'Olympus' -sources

Amazon (AMZN.O) is investing millions in training an ambitious large language model (LLMs), hoping it could rival top models from OpenAI and Alphabet (GOOGL.O), two people familiar with the matter told Reuters.

The model, codenamed as “Olympus”, has 2 trillion parameters, the people said, which could make it one of the largest models being trained. OpenAI's GPT-4 model, one of the best models available, is reported to have one trillion parameters....................

 

[China]

After OpenAI’s stunning ouster, policymakers are warning that AI could turn the economy on its head

 

Consumer Financial Protection Bureau Director Rohit Chopra is warning that the nature of the technology powering artificial intelligence could lead to oligopolies with sweeping implications for the economy.

 

Chopra’s comments came in the wake of the stunning ouster of OpenAI’s co-founder and CEO Sam Altman. The reaction to Altman’s firing — and his subsequent hiring by Microsoft, OpenAI’s largest investor — underscores how AI’s future will likely rest with a handful of very large companies and their boards, Chopra told POLITICO on Monday.

 

“There is a race to develop the foundational AI models. There will probably not be tons of those models. It may in fact be a natural oligopoly,” said Chopra, a longtime critic of how tech companies have addressed privacy and competition issues.

 

AI programs will likely have applications across sectors, Chopra said, and “the fact that Big Tech companies are now overlapping with the major foundational AI models is adding even more questions about what we do to make sure that they do not have outsized power,” he added.

 

He’s not alone in his concern. For months, policymakers from the White House to the Securities and Exchange Commission have been strategizing over how to address some of the headaches AI could cause for financial institutions and markets.

 

Click on the link for the full article

[ixcuincle]

wait wtf

 

https://www.pcmag.com/news/sam-altman-is-returning-to-openai-as-ceo

 

 

[Cooked Crack]

 

[Cooked Crack]

 

[Califan007 The Constipated]

 

 

 

 

[China]

Think tank tied to tech billionaires played key role in Biden’s AI order

 

The RAND Corporation — a prominent international think tank that has recently been tied to a growing influence network backed by tech billionaires — played a key role in drafting President Joe Biden’s new executive order on artificial intelligence, according to an AI researcher with knowledge of the order’s drafting and a recording of an internal RAND meeting obtained by POLITICO.

 

The provisions advanced by RAND in the October executive order included a sweeping set of reporting requirements placed on the most powerful AI systems, ostensibly designed to lessen the technology’s catastrophic risks. Those requirements hew closely to the policy priorities pursued by Open Philanthropy, a group that pumped over $15 million into RAND this year.

 

Financed by billionaire Facebook co-founder and Asana CEO Dustin Moskovitz and his wife Cari Tuna, Open Philanthropy is a major funder of causes associated with “effective altruism” — an ideology, made famous by disgraced FTX founder Sam Bankman-Fried, that emphasizes a data-driven approach to philanthropy.

 

Effective altruists now focus much of their attention on AI and are increasingly pushing Washington to address the technology’s apocalyptic potential, including the risk that advanced AIs could one day be used to develop bioweapons. Critics claim that the movement’s focus on speculative future risks serves the interests of top tech companies by distracting policymakers from existing AI harms, including its tendency to promote racial bias or undermine copyright protections.

 

Click on the link for the full article

[Captain Wiggles]

The RAND Corporation.

 

 

Ah who could forget all the nice wars they helped consult us on. 🤣