A FAKE text generator funded by scandal-hit billionaire Elon Musk has been branded so dangerous that it must be locked away to keep it from the prying hands of cyber criminals.

Sheepish tech whizzes behind the tool, dubbed GPT-2, admitted the tech risked being "misused" if it were released from their secretive California lab.

The software can write news stories and works of fiction – dubbed "deepfakes for text" – by reading paragraphs of source material.

In the wrong hands, it could be used to impersonate innocent people, or create unstoppable quantities of fake news.

Experts at OpenAI, an experimental non-profit research group bankrolled by Musk, have vowed to keep their creation secret to avoid a global catastrophe.

Normally, the group releases its full research to the public.

But in this case it is publishing a restricted version while it weighs up the potential consequences of a full handout.

"Due to concerns about large language models being used to generate deceptive, biased, or abusive language at scale, we are only releasing a much smaller version," the firm said.

"We are not releasing the dataset, training code, or GPT-2 model weights."

Tech whizzes trained the AI to write by feeding it 8million web pages.

It learned the typical patterns used in written work and can now conjure its own frighteningly real text.

In a blog post, OpenAI provided examples of the prose the software generated.

Here is the human-written text they fed it: "In a shocking finding, scientist discovered a herd of unicorns living in a remote, previously unexplored valley, in the Andes Mountains. Even more surprising to the researchers was the fact that the unicorns spoke perfect English."

GPT-2 then finished off the piece, including its own fake quotes.

It wrote: "The scientist named the population, after their distinctive horn, Ovid’s Unicorn. These four-horned, silver-white unicorns were previously unknown to science.

"Now, after almost two centuries, the mystery of what sparked this odd phenomenon is finally solved.

"Dr. Jorge Pérez, an evolutionary biologist from the University of La Paz, and several companions, were exploring the Andes Mountains when they found a small valley, with no other animals or humans. Pérez noticed that the valley had what appeared to be a natural fountain, surrounded by two peaks of rock and silver snow."

The tool took 10 attempts to come up with the text, according to OpenAI.

Its boffins said government regulation might be needed in future to ensure AI tech progresses safely.

"Governments should consider expanding or commencing initiatives to more systematically monitor the societal impact and diffusion of AI technologies, and to measure the progression in the capabilities of such systems," they said.

Elon Musk, who heads up Tesla and private rocket firm SpaceX, has previously warned of AI's dangerous potential.

He claims the performance of humans' puny brains will be outmatched by computers within the next 11 years.

Despite his hand in a wide range of experiment tech companies, the 47-year-old has vowed not to build killer robots over fears they could "destabilise every country on Earth".

In fact, he wants to save the human race from a robot rebellion by wiring our brains to the internet to make us all geniuses.

What do you think OpenAI should do with their tool? Let us know in the comments!

Originally Posted by starrdust

A FAKE text generator funded by scandal-hit billionaire Elon Musk has been branded so dangerous that it must be locked away to keep it from the prying hands of cyber criminals.

Sheepish tech whizzes behind the tool, dubbed GPT-2, admitted the tech risked being "misused" if it were released from their secretive California lab.

The software can write news stories and works of fiction – dubbed "deepfakes for text" – by reading paragraphs of source material.

In the wrong hands, it could be used to impersonate innocent people, or create unstoppable quantities of fake news.

Experts at OpenAI, an experimental non-profit research group bankrolled by Musk, have vowed to keep their creation secret to avoid a global catastrophe.

Normally, the group releases its full research to the public.

But in this case it is publishing a restricted version while it weighs up the potential consequences of a full handout.

"Due to concerns about large language models being used to generate deceptive, biased, or abusive language at scale, we are only releasing a much smaller version," the firm said.

"We are not releasing the dataset, training code, or GPT-2 model weights."

Tech whizzes trained the AI to write by feeding it 8million web pages.

It learned the typical patterns used in written work and can now conjure its own frighteningly real text.

In a blog post, OpenAI provided examples of the prose the software generated.

Here is the human-written text they fed it: "In a shocking finding, scientist discovered a herd of unicorns living in a remote, previously unexplored valley, in the Andes Mountains. Even more surprising to the researchers was the fact that the unicorns spoke perfect English."

GPT-2 then finished off the piece, including its own fake quotes.

It wrote: "The scientist named the population, after their distinctive horn, Ovid’s Unicorn. These four-horned, silver-white unicorns were previously unknown to science.

"Now, after almost two centuries, the mystery of what sparked this odd phenomenon is finally solved.

"Dr. Jorge Pérez, an evolutionary biologist from the University of La Paz, and several companions, were exploring the Andes Mountains when they found a small valley, with no other animals or humans. Pérez noticed that the valley had what appeared to be a natural fountain, surrounded by two peaks of rock and silver snow."

The tool took 10 attempts to come up with the text, according to OpenAI.

Its boffins said government regulation might be needed in future to ensure AI tech progresses safely.

"Governments should consider expanding or commencing initiatives to more systematically monitor the societal impact and diffusion of AI technologies, and to measure the progression in the capabilities of such systems," they said.

Elon Musk, who heads up Tesla and private rocket firm SpaceX, has previously warned of AI's dangerous potential.

He claims the performance of humans' puny brains will be outmatched by computers within the next 11 years.

Despite his hand in a wide range of experiment tech companies, the 47-year-old has vowed not to build killer robots over fears they could "destabilise every country on Earth".

In fact, he wants to save the human race from a robot rebellion by wiring our brains to the internet to make us all geniuses.

What do you think OpenAI should do with their tool? Let us know in the comments!

I am still more frightened of politicians rather than this bot.